diff --git a/README.md b/README.md
index ff23990..aa0b10e 100644
--- a/README.md
+++ b/README.md
@@ -373,7 +373,8 @@ perfkit checks the steps in order:
 
 1. **Is the measurement valid?** A step is invalid when the load generator, not the server, was
    the bottleneck: SIPp used more than 80% of one core, placed less than 95% of the offered rate,
-   or the load generator dropped UDP packets. Then the verdict is **inconclusive**, the reason
+   placed more than 105% of it (a burst of calls it owed from earlier steps), or the load
+   generator dropped UDP packets. Then the verdict is **inconclusive**, the reason
    names the step and the cause, and no limit is published.
 2. **Does it pass?** A step passes when p99 is **below** `p99LatencyMs` and failed % is **below**
    `failedCallsPct`. Both limits are strict: a p99 of exactly 50 ms breaks a 50 ms limit. The
diff --git a/package-lock.json b/package-lock.json
index 6fdee7b..9712281 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -1,12 +1,12 @@
 {
   "name": "perfkit",
-  "version": "0.1.6",
+  "version": "0.1.7",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "perfkit",
-      "version": "0.1.6",
+      "version": "0.1.7",
       "license": "MIT",
       "dependencies": {
         "commander": "^14.0.3",
diff --git a/package.json b/package.json
index aab21aa..8c6adc9 100644
--- a/package.json
+++ b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "perfkit",
-  "version": "0.1.6",
+  "version": "0.1.7",
   "description": "Benchmark harness for Fonoster projects: DigitalOcean runs, one report.json schema, one branded report page",
   "license": "MIT",
   "author": "Pedro Sanders <psanders@fonoster.com>",
diff --git a/src/analyze/measureCapacityStep.ts b/src/analyze/measureCapacityStep.ts
index 0d3ec64..e3865f7 100644
--- a/src/analyze/measureCapacityStep.ts
+++ b/src/analyze/measureCapacityStep.ts
@@ -19,6 +19,11 @@ import { udpDropsOver } from "./udpDropsOver.js";
 export const MAX_SIPP_CPU_PCT = 80;
 /** Placing less than this share of the offered rate means SIPp fell behind. */
 export const MIN_ACHIEVED_RATIO = 0.95;
+/**
+ * Placing more than this share means SIPp caught up on calls it owed from
+ * earlier steps in a burst, so the step measures the burst.
+ */
+export const MAX_ACHIEVED_RATIO = 1.05;
 /** The load generator's SIPp process, as the sampler names it. */
 export const SIPP_UAC_PROCESS = "sipp-uac";
 
@@ -43,7 +48,7 @@ export interface MeasuredStep {
  * Measures one capacity step. Latency counts the calls whose request was
  * sent in the step; attempts, failures and retransmissions come from the
  * stat rows in it. A step is invalid when SIPp used more than 80% of one
- * core, placed less than 95% of the offered rate, or the load generator's
+ * core, placed less than 95% or more than 105% of the offered rate, or the load generator's
  * UDP drops rose.
  */
 export function measureCapacityStep(input: StepInput): MeasuredStep {
@@ -78,9 +83,11 @@ export function measureCapacityStep(input: StepInput): MeasuredStep {
       ? `load generator saturated at ${rate} (SIPp at ${Math.round(sippCpuPct)}% of one core)`
       : achievedCps < offeredCps * MIN_ACHIEVED_RATIO
         ? `load generator fell behind at ${rate} (placed ${formatNumber(achievedCps)} ${input.unit}/s, ${Math.round((achievedCps / offeredCps) * 100)}% of offered)`
-        : loadgenUdpDrops > 0
-          ? `load generator dropped UDP packets at ${rate} (${formatNumber(loadgenUdpDrops)} buffer errors)`
-          : null;
+        : achievedCps > offeredCps * MAX_ACHIEVED_RATIO
+          ? `load generator sent a burst at ${rate} (placed ${formatNumber(achievedCps)} ${input.unit}/s, ${Math.round((achievedCps / offeredCps) * 100)}% of offered)`
+          : loadgenUdpDrops > 0
+            ? `load generator dropped UDP packets at ${rate} (${formatNumber(loadgenUdpDrops)} buffer errors)`
+            : null;
   return {
     failedPct,
     invalidReason,
diff --git a/test/analyze/createAnalyzeCapacity.unit.test.ts b/test/analyze/createAnalyzeCapacity.unit.test.ts
index 1fb58d3..c270b54 100644
--- a/test/analyze/createAnalyzeCapacity.unit.test.ts
+++ b/test/analyze/createAnalyzeCapacity.unit.test.ts
@@ -241,6 +241,23 @@ describe("createAnalyzeCapacity", () => {
       expect(result).not.to.have.property("rules");
     });
 
+    it("should be inconclusive when SIPp placed 110% of the offered rate", async () => {
+      // SIPp that falls behind its schedule later sends the calls it owes in a
+      // burst; that step measures the burst, not the server under test
+      // Act
+      const result = await analyze([
+        { offeredCps: 50 },
+        { offeredCps: 100, attempted: 6600 }
+      ]);
+
+      // Assert
+      expect(result).to.include({
+        verdict: "inconclusive",
+        reason:
+          "load generator sent a burst at 100 messages/s (placed 110 messages/s, 110% of offered)"
+      });
+    });
+
     it("should be inconclusive when SIPp placed 90% of the offered rate", async () => {
       // Act
       const result = await analyze([