Steam-Assisted Gravity Drainage (SAGD) is the leading recovering technique for heavy oil and bitumen in Canada. An indispensable condition for a successful SAGD process is the full utilization of the injected steam. However, because of the reservoir heterogeneity and the wellbore hydraulics and undulation, a non-uniform steam chamber will evolve. This study proposes Transient Temperature Analysis (TTA) to evaluate the earlystage SAGD (start-up and ramp-up) steam conformance through analyzing the temperature falloff data after shut-in. Both forward and inverse mathematical models are presented to facilitate the application of this technique. The four 2D forward mathematic models, including three non-condensation models of a two-system (a hot-zone of steam temperature and a cold-zone of reservoir temperature), a three-system (a transition-zone is added in between), superposition of multiple two-systems for irregular temperature distribution, and one condensation model, were developed to model the temperature falloff behaviour. All three non-condensation models assume that heat conduction is dominant after shut-in, whereas the condensation model incorporates the effect of steam condensation. The inverse model helps to automatically analyze the temperature falloff data and to interpret the hot-zone size. Sensitivity analyses show that the hot-zone size and shape, the observing distance, and the thermal diffusivity in the vicinity of the observing location (usually at the producer) strongly affect the temperature response. Synthetic case studies suggest that both the superposition model and three-system model can be used to interpret the temperature distribution for a SAGD start-up process under different conditions; the superposition model is applied before steam breakthrough; and the three-system model is applied after steam-breakthrough. Synthetic case studies also show that the condensation model can reasonably interpret the steam chamber size and match the temperature falloff for the ramp-up stage. The second part of this study investigated the effects of reservoir heterogeneity on the steam chamber development and temperature response after shut-in through 3D heterogeneous reservoir simulation studies. Based on the statistics gathered from the simulations, empirical correlation(s) between steam chamber size and the temperature falloff data was achieved through regression. A practical guideline regarding how to utilize these empirical correlations and to estimate the steam chamber size is introduced.