diff --git a/sentry_sdk/_span_batcher.py b/sentry_sdk/_span_batcher.py
index c1c40044ee..2033cf7845 100644
--- a/sentry_sdk/_span_batcher.py
+++ b/sentry_sdk/_span_batcher.py
@@ -30,6 +30,7 @@ class SpanBatcher(Batcher["SpanJSON"]):
     GLOBAL_MAX_BEFORE_DROP = 10_000
 
     MAX_BYTES_BEFORE_FLUSH = 5 * 1024 * 1024  # 5 MB
+    GLOBAL_MAX_BYTES_BEFORE_FLUSH = 25 * 1024 * 1024  # 25 MB
 
     FLUSH_WAIT_TIME = 5.0
 
@@ -50,6 +51,8 @@ def __init__(
         self._span_number: int = 0
 
         self._running_size: dict[str, int] = defaultdict(lambda: 0)
+        self._total_running_size: int = 0
+
         self._capture_func = capture_func
         self._record_lost_func = record_lost_func
         self._running = True
@@ -79,6 +82,8 @@ def _reset_thread_state(self) -> None:
         self._span_number = 0
 
         self._running_size = defaultdict(lambda: 0)
+        self._total_running_size = 0
+
         self._running = True
 
         self._lock = threading.Lock()
@@ -100,7 +105,7 @@ def _flush_loop(self) -> None:
 
             self._flush(only_pending=True)
 
-            if (
+            if self._total_running_size >= self.GLOBAL_MAX_BYTES_BEFORE_FLUSH or (
                 time.monotonic() - self._last_full_flush
                 >= self.FLUSH_WAIT_TIME + jitter
             ):
@@ -137,7 +142,9 @@ def add(self, span: "SpanJSON") -> None:
                 self._span_buffer[span["trace_id"]].append(span)
                 self._span_number += 1
 
-                self._running_size[span["trace_id"]] += self._estimate_size(span)
+                estimated_size = self._estimate_size(span)
+                self._running_size[span["trace_id"]] += estimated_size
+                self._total_running_size += estimated_size
 
                 if (
                     len(self._span_buffer[span["trace_id"]]) >= self.MAX_BEFORE_FLUSH
@@ -147,7 +154,9 @@ def add(self, span: "SpanJSON") -> None:
                     self._pending_flush.add(span["trace_id"])
                     notify = True
                 else:
-                    notify = False
+                    notify = (
+                        self._total_running_size >= self.GLOBAL_MAX_BYTES_BEFORE_FLUSH
+                    )
 
             if notify:
                 self._flush_event.set()
@@ -241,6 +250,7 @@ def _flush(self, only_pending: bool = False) -> None:
                 self._span_number -= len(self._span_buffer[bucket_id])
                 del self._span_buffer[bucket_id]
 
+                self._total_running_size -= self._running_size[bucket_id]
                 del self._running_size[bucket_id]
 
         for envelope in envelopes:
diff --git a/tests/tracing/test_span_batcher.py b/tests/tracing/test_span_batcher.py
index 507fef184b..fa34ab26e1 100644
--- a/tests/tracing/test_span_batcher.py
+++ b/tests/tracing/test_span_batcher.py
@@ -338,6 +338,69 @@ def test_weight_based_flushing_by_attribute_size(
     assert envelopes[0].items[0].payload.json["items"][1]["name"] == "big span"
 
 
+def test_global_length_based_flushing(sentry_init, capture_items, monkeypatch):
+    """When the batcher reaches GLOBAL_MAX_BYTES_BEFORE_FLUSH, all buckets will be flushed."""
+    # Limit of 2_000 is just above the size of a bare span.
+    monkeypatch.setattr(SpanBatcher, "GLOBAL_MAX_BYTES_BEFORE_FLUSH", 2_000)
+    # set the time-based flush limit to something huge so that it doesn't
+    # interfere
+    monkeypatch.setattr(SpanBatcher, "FLUSH_WAIT_TIME", 100000)
+
+    sentry_init(
+        traces_sample_rate=1.0,
+        trace_lifecycle="stream",
+    )
+
+    items = capture_items("span")
+
+    with sentry_sdk.traces.start_span(name="span"):
+        pass
+
+    sentry_sdk.traces.new_trace()
+    with sentry_sdk.traces.start_span(name="span"):
+        pass
+
+    time.sleep(0.1)
+
+    assert len(items) == 2
+    assert items[0].payload["name"] == "span"
+
+
+def test_total_size_reset_after_length_based_flushing(
+    sentry_init, capture_items, monkeypatch
+):
+    """Span is not flushed after a flush reduces the combined span size in bytes below the global limit."""
+    # Limit of 2_000 is just above the size of a bare span.
+    monkeypatch.setattr(SpanBatcher, "GLOBAL_MAX_BYTES_BEFORE_FLUSH", 2_000)
+    # set the time-based flush limit to something huge so that it doesn't
+    # interfere
+    monkeypatch.setattr(SpanBatcher, "FLUSH_WAIT_TIME", 100000)
+
+    sentry_init(
+        traces_sample_rate=1.0,
+        trace_lifecycle="stream",
+    )
+
+    items = capture_items("span")
+
+    with sentry_sdk.traces.start_span(name="span"):
+        pass
+
+    sentry_sdk.traces.new_trace()
+    with sentry_sdk.traces.start_span(name="span"):
+        pass
+
+    time.sleep(0.1)
+
+    with sentry_sdk.traces.start_span(name="span"):
+        pass
+
+    time.sleep(0.1)
+
+    assert len(items) == 2
+    assert items[0].payload["name"] == "span"
+
+
 def test_bucket_recreated_after_flush(sentry_init, capture_envelopes, monkeypatch):
     """Spans for a trace that arrive after that trace's bucket was flushed land in a fresh bucket."""
     monkeypatch.setattr(SpanBatcher, "MAX_BEFORE_FLUSH", 2)
@@ -545,6 +608,8 @@ def test_span_batcher_lock_reset_in_child_after_fork(sentry_init):
     batcher._span_number = 1
 
     batcher._running_size["test-trace-id"] = 42
+    batcher._total_running_size = 42
+
     batcher._active.flag = True
     batcher._flush_event.set()
     batcher._running = False
@@ -559,6 +624,7 @@ def test_span_batcher_lock_reset_in_child_after_fork(sentry_init):
         span_number_reset = batcher._span_number == 0
 
         running_size_reset = len(batcher._running_size) == 0
+        total_running_size_reset = batcher._total_running_size == 0
 
         active_reset = not getattr(batcher._active, "flag", False)
         event_reset = not batcher._flush_event.is_set()
@@ -572,6 +638,7 @@ def test_span_batcher_lock_reset_in_child_after_fork(sentry_init):
             and span_buffer_reset
             and span_number_reset
             and running_size_reset
+            and total_running_size_reset
             and active_reset
             and event_reset
             and running_reset