diff --git a/modules/core/src/norm.cpp b/modules/core/src/norm.cpp
index bbefefc95d2fec8d398138d98602a1eb7f94296b..4df25f495722987ba1b0952d0bfebcd31f90324b 100644
--- a/modules/core/src/norm.cpp
+++ b/modules/core/src/norm.cpp
@@ -1194,7 +1194,7 @@ double norm( InputArray _src1, InputArray _src2, int normType, InputArray _mask
         // special case to handle "integer" overflow in accumulator
         const size_t esz = src1.elemSize();
         const int total = (int)it.size;
-        const int intSumBlockSize = normType == NORM_L1 && depth <= CV_8S ? (1 << 23) : (1 << 15);
+        const int intSumBlockSize = (normType == NORM_L1 && depth <= CV_8S ? (1 << 23) : (1 << 15))/cn;
         const int blockSize = std::min(total, intSumBlockSize);
         int isum = 0;
         int count = 0;
diff --git a/modules/core/test/test_arithm.cpp b/modules/core/test/test_arithm.cpp
index effb0e68e05b3cc5c876a47e5f30547515330403..9e8e242d604a37c3378a65cfe723ab52cc6d4b4b 100644
--- a/modules/core/test/test_arithm.cpp
+++ b/modules/core/test/test_arithm.cpp
@@ -2166,6 +2166,15 @@ TEST(Core_Norm, IPP_regression_NORM_L1_16UC3_small)
     EXPECT_EQ((double)20*cn, cv::norm(a, b, NORM_L1, mask));
 }
 
+TEST(Core_Norm, NORM_L2_8UC4)
+{
+    // Tests there is no integer overflow in norm computation for multiple channels.
+    const int kSide = 100;
+    cv::Mat4b a(kSide, kSide, cv::Scalar(255, 255, 255, 255));
+    cv::Mat4b b = cv::Mat4b::zeros(kSide, kSide);
+    const double kNorm = 2.*kSide*255.;
+    EXPECT_EQ(kNorm, cv::norm(a, b, NORM_L2));
+}
 
 TEST(Core_ConvertTo, regression_12121)
 {