diff --git a/CHANGES.md b/CHANGES.md
index 6b6327c1..e02a1368 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -15,6 +15,9 @@ development source code and as such may not be routinely kept up to date.
 
 ## Bug fixes
 
+* `nextstrain build --aws-batch` now waits 10 minutes before automatically
+  canceling any jobs that are stuck as RUNNABLE due to misconfiguration.
+  ([#551](https://github.com/nextstrain/cli/pull/551))
 * Commands that authenticate with nextstrain.org, such as `nextstrain login`,
   `nextstrain whoami`, and `nextstrain remote` with nextstrain.org addresses,
   no longer load AWS credentials, which they don't use.  Previously, a local
diff --git a/doc/changes.md b/doc/changes.md
index 01f61c73..959c9621 100644
--- a/doc/changes.md
+++ b/doc/changes.md
@@ -19,6 +19,9 @@ development source code and as such may not be routinely kept up to date.
 (v-next-bug-fixes)=
 ### Bug fixes
 
+* `nextstrain build --aws-batch` now waits 10 minutes before automatically
+  canceling any jobs that are stuck as RUNNABLE due to misconfiguration.
+  ([#551](https://github.com/nextstrain/cli/pull/551))
 * Commands that authenticate with nextstrain.org, such as `nextstrain login`,
   `nextstrain whoami`, and `nextstrain remote` with nextstrain.org addresses,
   no longer load AWS credentials, which they don't use.  Previously, a local
diff --git a/nextstrain/cli/runner/aws_batch/__init__.py b/nextstrain/cli/runner/aws_batch/__init__.py
index 011cacb0..07c94456 100644
--- a/nextstrain/cli/runner/aws_batch/__init__.py
+++ b/nextstrain/cli/runner/aws_batch/__init__.py
@@ -420,11 +420,15 @@ def interrupt_signaled(sig, frame):
         if job.status_changed and not job.is_complete:
             print_stage("Job now %s" % job.status)
 
-        # Cancel jobs that are stuck due to misconfiguration.
-        # <https://docs.aws.amazon.com/batch/latest/userguide/job_stuck_in_runnable_misconfiguration.html>
+        # Cancel jobs that are stuck due to misconfiguration.¹ The 10 minute
+        # timeout is buffer time for EC2 instance scale-up, etc. and mirrors the
+        # minimum duration of the AWS-side equivalent.²
+        # ¹ <https://docs.aws.amazon.com/batch/latest/userguide/job_stuck_in_runnable_misconfiguration.html>
+        # ² <https://docs.aws.amazon.com/batch/latest/userguide/job_stuck_in_runnable_time_limit_actions.html>
         if (job.status == "RUNNABLE"
         and job.status_reason is not None
         and job.status_reason.startswith("MISCONFIGURATION")
+        and job.elapsed_time >= 600
         and not job.stop_sent):
             warn(f"Job is stuck in RUNNABLE. Reason: {job.status_reason}")
             print_stage("Automatically canceling stuck job…")
diff --git a/nextstrain/cli/runner/aws_batch/jobs.py b/nextstrain/cli/runner/aws_batch/jobs.py
index 9901d6e6..97bd8c6a 100644
--- a/nextstrain/cli/runner/aws_batch/jobs.py
+++ b/nextstrain/cli/runner/aws_batch/jobs.py
@@ -95,9 +95,9 @@ def is_failed(self) -> bool:
     @property
     def elapsed_time(self) -> float:
         # Timestamps from the job state are in milliseconds
-        created = self.state.get("createdAt", float("NaN"))  / 1000
-        stopped = self.state.get("stoppedAt", time() * 1000) / 1000
-        return stopped - created
+        start = self.state.get("createdAt", float("NaN"))  / 1000
+        end   = self.state.get("stoppedAt", time() * 1000) / 1000
+        return end - start
 
     @property
     def log_stream(self) -> Optional[str]: