diff --git a/studio/src/lib/csv.js b/studio/src/lib/csv.js
index 55ddfb8..fc49527 100644
--- a/studio/src/lib/csv.js
+++ b/studio/src/lib/csv.js
@@ -48,11 +48,18 @@ export function normalize(row) {
   };
 }
 
+// Above this size, parse on a background worker so a big paste/upload doesn't
+// freeze the UI. Small inputs parse on the main thread — the worker's
+// serialization overhead makes it slower for them, and PapaParse falls back to
+// synchronous parsing anyway when no Worker global exists (e.g. in tests).
+const WORKER_THRESHOLD_BYTES = 512 * 1024;
+
 export function parseCSV(text) {
   return new Promise((resolve, reject) => {
     Papa.parse(text, {
       header: true,
       skipEmptyLines: true,
+      worker: text.length > WORKER_THRESHOLD_BYTES,
       complete: (res) => {
         const rows = (res.data || [])
           .map(normalize)
diff --git a/studio/src/lib/csv.test.js b/studio/src/lib/csv.test.js
index 4c072bd..92a47f1 100644
--- a/studio/src/lib/csv.test.js
+++ b/studio/src/lib/csv.test.js
@@ -84,4 +84,15 @@ describe("parseCSV", () => {
     const rows = await parseCSV(csv);
     expect(rows[0].uid).toBe("999");
   });
+
+  it("parses a large input (past the worker threshold) correctly", async () => {
+    const lines = ["text,likeCount"];
+    for (let i = 0; i < 30000; i++) lines.push(`tweet number ${i},${i}`);
+    const csv = lines.join("\n");
+    expect(csv.length).toBeGreaterThan(512 * 1024); // exercises worker:true path
+    const rows = await parseCSV(csv);
+    expect(rows).toHaveLength(30000);
+    expect(rows[0].text).toBe("tweet number 0");
+    expect(rows[29999].likes).toBe(29999);
+  });
 });