The MITgcm (MIT general circulation model) has been applied to study geophysical flows with a broad range of scales. However, in the majority of applications, the accuracy of model results remains uncertain as the model has not been validated against field observations. In this paper, temperature measurements from Cayuga Lake are used to validate the MITgcm and study the sensitivity of the model to different model parameters. Both quantitative and qualitative methods are used to assess model skill. It is found that the linear equation of state (EOS) yields poorer results in comparison to polynomial EOS formulations. The vertical stratification shows a strong sensitivity to the background vertical viscosity when the KPP mixing scheme is employed. It is shown that the root mean square error between modeled and observed data is 1.9°C minimized and the observed vertical stratification is reproduced well with a Smagorinsky scheme for the horizontal eddy viscosity and a background vertical eddy viscosity of 10⁻³ m² s⁻¹, with a polynomial EOS.