Heat transfer to and from a circular cylinder in a cross-flow of water at low Reynolds number was studied both experimentally and numerically. The experiments were carried out in a high aspect ratio water channel. The test section inflow temperature and velocity, channel lower surface temperature and cylinder surface temperature were controlled to yield either laminar or turbulent flow for a desired Richardson number. When the lower surface was unheated, the temperatures of the lower surface and water upstream of the cylinder were maintained approximately equal and the flow was laminar. When the lower surface was heated, turbulence intensities as high as 20% were measured several cylinder diameters upstream of the cylinder due to turbulent thermal plumes produced by heating the lower surface. Variable property, two-dimensional simulations were undertaken using a variant of the v²–f turbulence model with buoyancy production of turbulence accounted for by a simple gradient diffusion model. Predicted and measured heat flux distributions around the cylinder are compared for values of the Richardson number, , from 0.3 to 9.3. For laminar flow, the predicted and measured heat flux results agreed to within the experimental uncertainty. When the lower surface was heated, and the flow was turbulent, there was qualitative agreement between predicted and measured heat flux distributions around the cylinder. However the predicted spatially averaged Nusselt number was from 37% to 53% larger than the measured spatially averaged Nusselt number. Additionally, spatially averaged Nusselt numbers are compared to correlations in the literature for mixed convection heat transfer to/from cylinders in cross-flow. The results presented here are larger than the correlation values. This is believed to be due to the effects of buoyancy-induced turbulence resulting from heating the lower surface and the proximity of the cylinder to that surface.