diff --git a/rclcpp/CMakeLists.txt b/rclcpp/CMakeLists.txt
index 81db9a365d..dd2bc51434 100644
--- a/rclcpp/CMakeLists.txt
+++ b/rclcpp/CMakeLists.txt
@@ -23,7 +23,7 @@ if(NOT CMAKE_CXX_STANDARD)
   set(CMAKE_CXX_STANDARD 14)
 endif()
 if(CMAKE_COMPILER_IS_GNUCXX OR CMAKE_CXX_COMPILER_ID MATCHES "Clang")
-  add_compile_options(-Wall -Wextra -Wpedantic -Wnon-virtual-dtor -Woverloaded-virtual)
+  add_compile_options(-Wall -Wextra -Wconversion -Wpedantic -Wnon-virtual-dtor -Woverloaded-virtual)
 endif()
 
 set(${PROJECT_NAME}_SRCS
diff --git a/rclcpp/src/rclcpp/node_interfaces/node_parameters.cpp b/rclcpp/src/rclcpp/node_interfaces/node_parameters.cpp
index f5d0aa7e64..412fba46c5 100644
--- a/rclcpp/src/rclcpp/node_interfaces/node_parameters.cpp
+++ b/rclcpp/src/rclcpp/node_interfaces/node_parameters.cpp
@@ -159,7 +159,7 @@ __lockless_has_parameter(
 // see https://en.cppreference.com/w/cpp/types/numeric_limits/epsilon
 RCLCPP_LOCAL
 bool
-__are_doubles_equal(double x, double y, size_t ulp = 100)
+__are_doubles_equal(double x, double y, double ulp = 100.0)
 {
   return std::abs(x - y) <= std::numeric_limits<double>::epsilon() * std::abs(x + y) * ulp;
 }
diff --git a/rclcpp/src/rclcpp/node_options.cpp b/rclcpp/src/rclcpp/node_options.cpp
index cac10e6685..43c401828f 100644
--- a/rclcpp/src/rclcpp/node_options.cpp
+++ b/rclcpp/src/rclcpp/node_options.cpp
@@ -338,7 +338,7 @@ NodeOptions::get_domain_id_from_env() const
   _dupenv_s(&ros_domain_id, &ros_domain_id_size, env_var);
 #endif
   if (ros_domain_id) {
-    uint32_t number = strtoul(ros_domain_id, NULL, 0);
+    uint32_t number = static_cast<uint32_t>(strtoul(ros_domain_id, NULL, 0));
     if (number == (std::numeric_limits<uint32_t>::max)()) {
 #ifdef _WIN32
       // free the ros_domain_id before throwing, if getenv was used on Windows
diff --git a/rclcpp/test/rclcpp/executors/test_multi_threaded_executor.cpp b/rclcpp/test/rclcpp/executors/test_multi_threaded_executor.cpp
index bfe2b25457..b6fd2c3fda 100644
--- a/rclcpp/test/rclcpp/executors/test_multi_threaded_executor.cpp
+++ b/rclcpp/test/rclcpp/executors/test_multi_threaded_executor.cpp
@@ -83,7 +83,7 @@ TEST_F(TestMultiThreadedExecutor, timer_over_take) {
 
       {
         std::lock_guard<std::mutex> lock(last_mutex);
-        double diff = std::abs((now - last).nanoseconds()) / 1.0e9;
+        double diff = static_cast<double>(std::abs((now - last).nanoseconds())) / 1.0e9;
         last = now;
 
         if (diff < PERIOD - TOLERANCE) {
diff --git a/rclcpp/test/rclcpp/test_time.cpp b/rclcpp/test/rclcpp/test_time.cpp
index 141064e13d..8df4b341d2 100644
--- a/rclcpp/test/rclcpp/test_time.cpp
+++ b/rclcpp/test/rclcpp/test_time.cpp
@@ -275,8 +275,8 @@ TEST(TestTime, overflow_detectors) {
   // 256 * 256 = 64K total loops, should be pretty fast on everything
   for (big_type_t y = min_val; y <= max_val; ++y) {
     for (big_type_t x = min_val; x <= max_val; ++x) {
-      const big_type_t sum = x + y;
-      const big_type_t diff = x - y;
+      const big_type_t sum = static_cast<big_type_t>(x + y);
+      const big_type_t diff = static_cast<big_type_t>(x - y);
 
       const bool add_will_overflow =
         rclcpp::add_will_overflow(test_type_t(x), test_type_t(y));