diff --git a/internal/database/database.go b/internal/database/database.go
index 06d611d..0cc7c4e 100644
--- a/internal/database/database.go
+++ b/internal/database/database.go
@@ -6,24 +6,32 @@
 // multiple source files. Blobs are content-addressed, meaning their filename
 // is derived from their SHA256 hash after compression and encryption.
 //
-// The database does not support migrations. If the schema changes, delete
-// the local database and perform a full backup to recreate it.
+// Schema is managed via numbered SQL migrations embedded in the schema/
+// directory. Migration 000.sql bootstraps the schema_migrations tracking
+// table; subsequent migrations (001, 002, …) are applied in order.
 package database
 
 import (
 	"context"
 	"database/sql"
-	_ "embed"
+	"embed"
 	"fmt"
 	"os"
+	"path/filepath"
+	"sort"
+	"strconv"
 	"strings"
 
 	"git.eeqj.de/sneak/vaultik/internal/log"
 	_ "modernc.org/sqlite"
 )
 
-//go:embed schema.sql
-var schemaSQL string
+//go:embed schema/*.sql
+var schemaFS embed.FS
+
+// bootstrapVersion is the migration that creates the schema_migrations
+// table itself. It is applied before the normal migration loop.
+const bootstrapVersion = 0
 
 // DB represents the Vaultik local index database connection.
 // It uses SQLite to track file metadata, content-defined chunks, and blob associations.
@@ -35,6 +43,46 @@ type DB struct {
 	path string
 }
 
+// ParseMigrationVersion extracts the numeric version prefix from a migration
+// filename. Filenames must follow the pattern "<version>.sql" or
+// "<version>_<description>.sql", where version is a zero-padded numeric
+// string (e.g. "001", "002"). Returns the version as an integer and an
+// error if the filename does not match the expected pattern.
+func ParseMigrationVersion(filename string) (int, error) {
+	name := strings.TrimSuffix(filename, filepath.Ext(filename))
+	if name == "" {
+		return 0, fmt.Errorf("invalid migration filename %q: empty name", filename)
+	}
+
+	// Split on underscore to separate version from description.
+	// If there's no underscore, the entire stem is the version.
+	versionStr := name
+	if idx := strings.IndexByte(name, '_'); idx >= 0 {
+		versionStr = name[:idx]
+	}
+
+	if versionStr == "" {
+		return 0, fmt.Errorf("invalid migration filename %q: empty version prefix", filename)
+	}
+
+	// Validate the version is purely numeric.
+	for _, ch := range versionStr {
+		if ch < '0' || ch > '9' {
+			return 0, fmt.Errorf(
+				"invalid migration filename %q: version %q contains non-numeric character %q",
+				filename, versionStr, string(ch),
+			)
+		}
+	}
+
+	version, err := strconv.Atoi(versionStr)
+	if err != nil {
+		return 0, fmt.Errorf("invalid migration filename %q: %w", filename, err)
+	}
+
+	return version, nil
+}
+
 // New creates a new database connection at the specified path.
 // It creates the schema if needed and configures SQLite with WAL mode for
 // better concurrency. SQLite handles crash recovery automatically when
@@ -72,9 +120,9 @@ func New(ctx context.Context, path string) (*DB, error) {
 			}
 
 			db := &DB{conn: conn, path: path}
-			if err := db.createSchema(ctx); err != nil {
+			if err := applyMigrations(ctx, conn); err != nil {
 				_ = conn.Close()
-				return nil, fmt.Errorf("creating schema: %w", err)
+				return nil, fmt.Errorf("applying migrations: %w", err)
 			}
 			return db, nil
 		}
@@ -125,9 +173,9 @@ func New(ctx context.Context, path string) (*DB, error) {
 	}
 
 	db := &DB{conn: conn, path: path}
-	if err := db.createSchema(ctx); err != nil {
+	if err := applyMigrations(ctx, conn); err != nil {
 		_ = conn.Close()
-		return nil, fmt.Errorf("creating schema: %w", err)
+		return nil, fmt.Errorf("applying migrations: %w", err)
 	}
 
 	log.Debug("Database connection established successfully", "path", path)
@@ -198,9 +246,120 @@ func (db *DB) QueryRowWithLog(
 	return db.conn.QueryRowContext(ctx, query, args...)
 }
 
-func (db *DB) createSchema(ctx context.Context) error {
-	_, err := db.conn.ExecContext(ctx, schemaSQL)
-	return err
+// collectMigrations reads the embedded schema directory and returns
+// migration filenames sorted lexicographically.
+func collectMigrations() ([]string, error) {
+	entries, err := schemaFS.ReadDir("schema")
+	if err != nil {
+		return nil, fmt.Errorf("failed to read schema directory: %w", err)
+	}
+
+	var migrations []string
+
+	for _, entry := range entries {
+		if !entry.IsDir() && strings.HasSuffix(entry.Name(), ".sql") {
+			migrations = append(migrations, entry.Name())
+		}
+	}
+
+	sort.Strings(migrations)
+
+	return migrations, nil
+}
+
+// bootstrapMigrationsTable ensures the schema_migrations table exists
+// by applying 000.sql if the table is missing.
+func bootstrapMigrationsTable(ctx context.Context, db *sql.DB) error {
+	var tableExists int
+
+	err := db.QueryRowContext(ctx,
+		"SELECT COUNT(*) FROM sqlite_master WHERE type='table' AND name='schema_migrations'",
+	).Scan(&tableExists)
+	if err != nil {
+		return fmt.Errorf("failed to check for migrations table: %w", err)
+	}
+
+	if tableExists > 0 {
+		return nil
+	}
+
+	content, err := schemaFS.ReadFile("schema/000.sql")
+	if err != nil {
+		return fmt.Errorf("failed to read bootstrap migration 000.sql: %w", err)
+	}
+
+	log.Info("applying bootstrap migration", "version", bootstrapVersion)
+
+	_, err = db.ExecContext(ctx, string(content))
+	if err != nil {
+		return fmt.Errorf("failed to apply bootstrap migration: %w", err)
+	}
+
+	return nil
+}
+
+// applyMigrations applies all pending migrations to db. It first bootstraps
+// the schema_migrations table via 000.sql, then iterates through remaining
+// migration files in order.
+func applyMigrations(ctx context.Context, db *sql.DB) error {
+	if err := bootstrapMigrationsTable(ctx, db); err != nil {
+		return err
+	}
+
+	migrations, err := collectMigrations()
+	if err != nil {
+		return err
+	}
+
+	for _, migration := range migrations {
+		version, parseErr := ParseMigrationVersion(migration)
+		if parseErr != nil {
+			return parseErr
+		}
+
+		// Check if already applied.
+		var count int
+
+		err := db.QueryRowContext(ctx,
+			"SELECT COUNT(*) FROM schema_migrations WHERE version = ?",
+			version,
+		).Scan(&count)
+		if err != nil {
+			return fmt.Errorf("failed to check migration status: %w", err)
+		}
+
+		if count > 0 {
+			log.Debug("migration already applied", "version", version)
+
+			continue
+		}
+
+		// Read and apply migration.
+		content, readErr := schemaFS.ReadFile(filepath.Join("schema", migration))
+		if readErr != nil {
+			return fmt.Errorf("failed to read migration %s: %w", migration, readErr)
+		}
+
+		log.Info("applying migration", "version", version)
+
+		_, execErr := db.ExecContext(ctx, string(content))
+		if execErr != nil {
+			return fmt.Errorf("failed to apply migration %s: %w", migration, execErr)
+		}
+
+		// Record migration as applied.
+		_, recErr := db.ExecContext(ctx,
+			"INSERT INTO schema_migrations (version) VALUES (?)",
+			version,
+		)
+		if recErr != nil {
+			return fmt.Errorf("failed to record migration %s: %w", migration, recErr)
+		}
+
+		log.Info("migration applied successfully", "version", version)
+	}
+
+	return nil
 }
 
 // NewTestDB creates an in-memory SQLite database for testing purposes.
diff --git a/internal/database/database_test.go b/internal/database/database_test.go
index 65457d1..6d763a3 100644
--- a/internal/database/database_test.go
+++ b/internal/database/database_test.go
@@ -2,6 +2,7 @@ package database
 
 import (
 	"context"
+	"database/sql"
 	"fmt"
 	"path/filepath"
 	"testing"
@@ -26,9 +27,10 @@ func TestDatabase(t *testing.T) {
 		t.Fatal("database connection is nil")
 	}
 
-	// Test schema creation (already done in New)
+	// Test schema creation (already done in New via migrations)
 	// Verify tables exist
 	tables := []string{
+		"schema_migrations",
 		"files", "file_chunks", "chunks", "blobs",
 		"blob_chunks", "chunk_files", "snapshots",
 	}
@@ -99,3 +101,139 @@ func TestDatabaseConcurrentAccess(t *testing.T) {
 		t.Errorf("expected 10 chunks, got %d", count)
 	}
 }
+
+func TestParseMigrationVersion(t *testing.T) {
+	tests := []struct {
+		name      string
+		filename  string
+		wantVer   int
+		wantError bool
+	}{
+		{name: "valid 000.sql", filename: "000.sql", wantVer: 0, wantError: false},
+		{name: "valid 001.sql", filename: "001.sql", wantVer: 1, wantError: false},
+		{name: "valid 099.sql", filename: "099.sql", wantVer: 99, wantError: false},
+		{name: "valid with description", filename: "001_initial_schema.sql", wantVer: 1, wantError: false},
+		{name: "valid large version", filename: "123_big_migration.sql", wantVer: 123, wantError: false},
+		{name: "invalid alpha version", filename: "abc.sql", wantVer: 0, wantError: true},
+		{name: "invalid mixed chars", filename: "12a.sql", wantVer: 0, wantError: true},
+		{name: "invalid no extension", filename: "schema.sql", wantVer: 0, wantError: true},
+		{name: "empty string", filename: "", wantVer: 0, wantError: true},
+	}
+
+	for _, tc := range tests {
+		t.Run(tc.name, func(t *testing.T) {
+			got, err := ParseMigrationVersion(tc.filename)
+			if tc.wantError {
+				if err == nil {
+					t.Errorf("ParseMigrationVersion(%q) = %d, nil; want error", tc.filename, got)
+				}
+				return
+			}
+			if err != nil {
+				t.Errorf("ParseMigrationVersion(%q) unexpected error: %v", tc.filename, err)
+				return
+			}
+			if got != tc.wantVer {
+				t.Errorf("ParseMigrationVersion(%q) = %d; want %d", tc.filename, got, tc.wantVer)
+			}
+		})
+	}
+}
+
+func TestApplyMigrations_Idempotent(t *testing.T) {
+	ctx := context.Background()
+
+	conn, err := sql.Open("sqlite", ":memory:?_foreign_keys=ON")
+	if err != nil {
+		t.Fatalf("failed to open database: %v", err)
+	}
+	defer func() {
+		if err := conn.Close(); err != nil {
+			t.Errorf("failed to close database: %v", err)
+		}
+	}()
+
+	conn.SetMaxOpenConns(1)
+	conn.SetMaxIdleConns(1)
+
+	// First run: apply all migrations.
+	if err := applyMigrations(ctx, conn); err != nil {
+		t.Fatalf("first applyMigrations failed: %v", err)
+	}
+
+	// Count rows in schema_migrations after first run.
+	var countBefore int
+	if err := conn.QueryRowContext(ctx, "SELECT COUNT(*) FROM schema_migrations").Scan(&countBefore); err != nil {
+		t.Fatalf("failed to count schema_migrations after first run: %v", err)
+	}
+
+	// Second run: must be a no-op.
+	if err := applyMigrations(ctx, conn); err != nil {
+		t.Fatalf("second applyMigrations failed: %v", err)
+	}
+
+	// Count rows in schema_migrations after second run — must be unchanged.
+	var countAfter int
+	if err := conn.QueryRowContext(ctx, "SELECT COUNT(*) FROM schema_migrations").Scan(&countAfter); err != nil {
+		t.Fatalf("failed to count schema_migrations after second run: %v", err)
+	}
+
+	if countBefore != countAfter {
+		t.Errorf("schema_migrations row count changed: before=%d, after=%d", countBefore, countAfter)
+	}
+}
+
+func TestBootstrapMigrationsTable_FreshDatabase(t *testing.T) {
+	ctx := context.Background()
+
+	conn, err := sql.Open("sqlite", ":memory:?_foreign_keys=ON")
+	if err != nil {
+		t.Fatalf("failed to open database: %v", err)
+	}
+	defer func() {
+		if err := conn.Close(); err != nil {
+			t.Errorf("failed to close database: %v", err)
+		}
+	}()
+
+	conn.SetMaxOpenConns(1)
+	conn.SetMaxIdleConns(1)
+
+	// Verify schema_migrations does NOT exist yet.
+	var tableBefore int
+	if err := conn.QueryRowContext(ctx,
+		"SELECT COUNT(*) FROM sqlite_master WHERE type='table' AND name='schema_migrations'",
+	).Scan(&tableBefore); err != nil {
+		t.Fatalf("failed to check for table before bootstrap: %v", err)
+	}
+	if tableBefore != 0 {
+		t.Fatal("schema_migrations table should not exist before bootstrap")
+	}
+
+	// Run bootstrap.
+	if err := bootstrapMigrationsTable(ctx, conn); err != nil {
+		t.Fatalf("bootstrapMigrationsTable failed: %v", err)
+	}
+
+	// Verify schema_migrations now exists.
+	var tableAfter int
+	if err := conn.QueryRowContext(ctx,
+		"SELECT COUNT(*) FROM sqlite_master WHERE type='table' AND name='schema_migrations'",
+	).Scan(&tableAfter); err != nil {
+		t.Fatalf("failed to check for table after bootstrap: %v", err)
+	}
+	if tableAfter != 1 {
+		t.Fatalf("schema_migrations table should exist after bootstrap, got count=%d", tableAfter)
+	}
+
+	// Verify version 0 row exists.
+	var version int
+	if err := conn.QueryRowContext(ctx,
+		"SELECT version FROM schema_migrations WHERE version = 0",
+	).Scan(&version); err != nil {
+		t.Fatalf("version 0 row not found in schema_migrations: %v", err)
+	}
+	if version != 0 {
+		t.Errorf("expected version 0, got %d", version)
+	}
+}
diff --git a/internal/database/schema/000.sql b/internal/database/schema/000.sql
new file mode 100644
index 0000000..e06a2da
--- /dev/null
+++ b/internal/database/schema/000.sql
@@ -0,0 +1,9 @@
+-- Migration 000: Schema migrations tracking table
+-- Applied as a bootstrap step before the normal migration loop.
+
+CREATE TABLE IF NOT EXISTS schema_migrations (
+    version INTEGER PRIMARY KEY,
+    applied_at DATETIME DEFAULT CURRENT_TIMESTAMP
+);
+
+INSERT OR IGNORE INTO schema_migrations (version) VALUES (0);
diff --git a/internal/database/schema.sql b/internal/database/schema/001.sql
similarity index 96%
rename from internal/database/schema.sql
rename to internal/database/schema/001.sql
index cdc8533..5f54565 100644
--- a/internal/database/schema.sql
+++ b/internal/database/schema/001.sql
@@ -1,6 +1,5 @@
--- Vaultik Database Schema
--- Note: This database does not support migrations. If the schema changes,
--- delete the local database and perform a full backup to recreate it.
+-- Migration 001: Initial Vaultik schema
+-- All core tables for tracking files, chunks, blobs, snapshots, and uploads.
 
 -- Files table: stores metadata about files in the filesystem
 CREATE TABLE IF NOT EXISTS files (
@@ -133,4 +132,4 @@ CREATE TABLE IF NOT EXISTS uploads (
 );
 
 -- Index for efficient snapshot lookups
-CREATE INDEX IF NOT EXISTS idx_uploads_snapshot_id ON uploads(snapshot_id);
\ No newline at end of file
+CREATE INDEX IF NOT EXISTS idx_uploads_snapshot_id ON uploads(snapshot_id);
diff --git a/internal/database/schema/008_uploads.sql b/internal/database/schema/008_uploads.sql
deleted file mode 100644
index 49b5add..0000000
--- a/internal/database/schema/008_uploads.sql
+++ /dev/null
@@ -1,11 +0,0 @@
--- Track blob upload metrics
-CREATE TABLE IF NOT EXISTS uploads (
-    blob_hash TEXT PRIMARY KEY,
-    uploaded_at TIMESTAMP NOT NULL,
-    size INTEGER NOT NULL,
-    duration_ms INTEGER NOT NULL,
-    FOREIGN KEY (blob_hash) REFERENCES blobs(blob_hash)
-);
-
-CREATE INDEX idx_uploads_uploaded_at ON uploads(uploaded_at);
-CREATE INDEX idx_uploads_duration ON uploads(duration_ms);
\ No newline at end of file