diff --git a/data_structures/heap/median_in_a_stream.py b/data_structures/heap/median_in_a_stream.py
new file mode 100644
index 000000000000..e534c5214c13
--- /dev/null
+++ b/data_structures/heap/median_in_a_stream.py
@@ -0,0 +1,104 @@
+"""
+Median in a stream using a heap-based approach.
+
+Reference:
+https://en.wikipedia.org/wiki/Median#Running_median
+"""
+
+import heapq
+
+
+def signum(a: int, b: int) -> int:
+    """
+    Return 1 if a > b, -1 if a < b, 0 if equal.
+    """
+    if a > b:
+        return 1
+    if a < b:
+        return -1
+    return 0
+
+
+def call_median(
+    element: int,
+    max_heap: list[int],
+    min_heap: list[int],
+    median: int,
+) -> int:
+    """
+    Update heaps and median based on the new element.
+
+    Args:
+        element (int): new element in stream
+        max_heap (list[int]): max heap (as negative numbers)
+        min_heap (list[int]): min heap
+        median (int): current median
+
+    Returns:
+        int: updated median
+    """
+    size_diff = signum(len(max_heap), len(min_heap))
+
+    if size_diff == 0:
+        if element > median:
+            heapq.heappush(min_heap, element)
+            median = min_heap[0]
+        else:
+            heapq.heappush(max_heap, -element)
+            median = -max_heap[0]
+    elif size_diff == 1:
+        if element > median:
+            heapq.heappush(min_heap, element)
+        else:
+            heapq.heappush(min_heap, -heapq.heappushpop(max_heap, -element))
+        median = (-max_heap[0] + min_heap[0]) // 2
+    else:  # size_diff == -1
+        if element > median:
+            heapq.heappush(max_heap, -heapq.heappushpop(min_heap, element))
+        else:
+            heapq.heappush(max_heap, -element)
+        median = (-max_heap[0] + min_heap[0]) // 2
+
+    return median
+
+
+def median_in_a_stream(arr: list[int]) -> list[int]:
+    """
+    Return the median after each new element in the stream.
+
+    Args:
+        arr (list[int]): list of integers
+
+    Returns:
+        list[int]: running medians
+
+    >>> median_in_a_stream([20, 14, 13, 16, 17])
+    [20, 17, 14, 15, 16]
+    >>> median_in_a_stream([5, 15, 1, 3])
+    [5, 10, 5, 4]
+    >>> median_in_a_stream([])
+    Traceback (most recent call last):
+    ...
+    ValueError: Input list must not be empty
+    """
+    if not arr:
+        raise ValueError("Input list must not be empty")
+
+    max_heap: list[int] = []  # left side (as negative numbers)
+    min_heap: list[int] = []  # right side
+    median = arr[0]
+    max_heap.append(-arr[0])
+    medians: list[int] = [median]
+
+    for element in arr[1:]:
+        median = call_median(element, max_heap, min_heap, median)
+        medians.append(median)
+
+    return medians
+
+
+if __name__ == "__main__":
+    n = int(input("Enter number of elements: ").strip())
+    arr = [int(input().strip()) for _ in range(n)]
+    result = median_in_a_stream(arr)
+    print("Running medians:", result)