Abstract Magnetic Resonance Thermometry (MRT) is a developing diagnostic technique that leverages advanced medical technologies to accurately measure the temperature of a fluid flow within and around complex geometries. The full three-dimensional temperature field obtained by MRT can be used to analyze heat transfer characteristics and potentially investigate thermal boundary layers near arbitrarily complex surfaces. This technique requires neither optical nor physical accessibility, thereby enabling a wide range of engineering applications. This paper describes the current state of the art for MRT measurement, detailing turbulent water channel tests, materials selection, scanning parameters, data analysis of time-averaged temperature measurements, and uncertainty estimates. The purpose of this work was to evaluate and refine the MRT technique to increase the accuracy of temperature measurements and minimize the error in fully turbulent flow measurements. In the present study, a plate with a vertical cylinder extending from both of its sides was placed between two channels, and a diagonal hole was drilled through the cylinder from one side of the plate to the other. This enabled fluid from one channel to mix with the fluid in the other. This experiment studied the mixing of two fluids at different temperatures. The upstream temperatures of each fluid were measured with thermocouples. Both flows were fully turbulent, and the colder temperature channel had a Reynolds number of 11,800. Tests were run with four different fluid temperatures for calibration and to determine any temperature dependence of measurements. Three-dimensional temperature field measurements are reported and details about data processing and procedures to conduct the experiments are provided. This work resulted in several notable improvements to MRT experimental methods. The test section and water channel were designed to limit the effects of thermal expansion in the stereolithography materials used for manufacturing the complex internal flow geometry. Multiple echo scans were used to minimize the effects of magnetic field drift commonly observed in extended scanning periods in MRI systems. Data analysis techniques were used to quantify expansion effects for both hot and cold flow cases. To quantify measurement uncertainty, the standard deviation of the mean was calculated at each data point across different scan numbers and confidence intervals established using a student t-test. An improved data processing code was used to filter data resulting in increased measurement accuracy and reduced uncertainty to less than 1 °C for most of the domain. Future work will further refine the experimental techniques to improve scanning procedures, employ high conductivity ceramics and larger geometries with relevant applications, and simplify data processing methods to generate full-field flow temperature data.