Miniaturization and integration of functional components on-chip offer great advantages. For many applications, the ability to control and measure temperature inside of microfluidic devices while in operation is critical as temperature affects biological processes and temperature variation perturbs such processes, e.g. embryo development. In addition, temperature variation can be a side-effect of some microfluidic techniques, e.g. Joule heating leading to hyperthermic cell damage. Thus, a reliable in situ temperature measurement method is required to design and operate microsystems effectively. Spectroscopic methods for temperature measurement in microfluidic devices have been developed, taking advantage of temperature-dependent properties of chemicals such as Rhodamine-B. However, there is still a need for both a simple inexpensive in situ measurement and with biological samples without toxicity effect from these dyes. We present for the first time 3D temperature measurements in microfluidic systems using Brownian motion of nanoparticles in the presence of live biological samples. The technique takes advantage of well defined thermal correlation of Brownian motion. Because the kinetic energy of small particles and fluid viscosity are functions of temperature, by measuring the Brownian diffusivity of suspended particles of known sizes, the surrounding liquid temperature can be determined. We measured mean square displacements (MSD) of 500-nm fluorescent particles for temperatures ranging from 5 °C to 50°C; the calculated temperature shows good agreement with the temperature measured by a thermocouple. Since tracking nanopaticles does not rely on fluorescence intensity, MSD measurements are not affected by the power of the incident light, spatial variation of excitation, concentration artifacts, and photobleaching. Therefore, this method is superior in its reproducibility and reduced systematic errors. To demonstrate the capability of this method as a 3D temperature mapping tool, we obtained Z-stack images of nanoparticles in a temperature controlled microfluidic channel. Nanoparticles out of focus can be easily discarded by MSD analysis algorithm, thereby achieving optical sectioning without expensive multiple-photon scanning setup. We showed temperature measurement with a resolution of ~ 4 ìm in the z-axis. The mapped 3D temperature profile in the channel was in good agreement with numerical simulations in COMSOL. Finally, we used this method for temperature measurement in the presence of biological samples.