diff --git a/test/promote_integral.cpp b/test/promote_integral.cpp
index 4178ef812..590d5571e 100644
--- a/test/promote_integral.cpp
+++ b/test/promote_integral.cpp
@@ -14,6 +14,7 @@
 //  - Remove support for boost::multiprecision types
 //  - Remove support for 128-bit integer types
 //  - Update and sort includes
+//  - Add explicit conversions to avoid warnings due to implicit integral promotions
 //
 #include <boost/gil/promote_integral.hpp>
 
@@ -46,7 +47,7 @@ struct absolute_value
 {
     static inline T apply(T const& t)
     {
-        return t < 0 ? -t : t;
+        return static_cast<T>(t < 0 ? -t : t);
     }
 };
 
@@ -73,11 +74,12 @@ struct test_max_values
         // but to avoid warning: comparing floating point with == is unsafe.
 
         Promoted min_value = (std::numeric_limits<Integral>::min)();
-        min_value *= min_value;
+        // Explicit casts to avoid warning: conversion to short int from int may alter its value
+        min_value = static_cast<Promoted>(min_value * min_value);
         BOOST_CHECK(absolute_value<Promoted>::apply(min_value) >= min_value);
         BOOST_CHECK(absolute_value<Promoted>::apply(min_value) <= min_value);
         Promoted max_value = (std::numeric_limits<Integral>::max)();
-        max_value *= max_value;
+        max_value = static_cast<Promoted>(max_value * max_value);
         BOOST_CHECK(absolute_value<Promoted>::apply(max_value) >= max_value);
         BOOST_CHECK(absolute_value<Promoted>::apply(max_value) <= max_value);
 
@@ -95,7 +97,7 @@ struct test_max_values<Integral, Promoted, false>
     static inline void apply()
     {
         Promoted max_value = (std::numeric_limits<Integral>::max)();
-        Promoted max_value_sqr = max_value * max_value;
+        Promoted max_value_sqr = static_cast<Promoted>(max_value * max_value);
         BOOST_CHECK(max_value_sqr < (std::numeric_limits<Promoted>::max)()
                     &&
                     max_value_sqr > max_value);