Radar-based latent heating retrievals typically apply a lookup table (LUT) derived from model output to surface rain amounts and rain type to determine the vertical structure of heating. In this study, a method has been developed that uses the size characteristics of precipitating systems (i.e., area and mean echo-top height) instead of rain amount to estimate latent heating profiles from radar observations. This technique [named the convectivestratiform area (CSA) algorithm] leverages the relationship between the organization of convective systems and the structure of latent heating profiles and avoids pitfalls associated with retrieving accurate rainfall information from radars and models. The CSA LUTs are based on a high-resolution regional model simulation over the equatorial Indian Ocean. The CSA LUTs show that convective latent heating increases in magnitude and height as area and echo-top heights grow, with a congestus signature of midlevel cooling for less vertically extensive convective systems. Stratiform latent heating varies weakly in vertical structure, but its magnitude is strongly linked to area and mean echo-top heights. The CSA LUT was applied to radar observations collected during the DYNAMO/Cooperative Indian Ocean Experiment on Intraseasonal Variability in the Year 2011 (CINDY2011)/ARM MJO Investigation Experiment (AMIE) field campaign, and the CSA heating retrieval was generally consistent with other measures of heating profiles. The impact of resolution and spatial mismatch between the model and radar grids is addressed, and unrealistic latent heating profiles in the stratiform LUT, namely, a low-level heating peak, an elevated melting layer, and net column cooling, were identified. These issues highlight the need for accurate convectivestratiform separations and improvement in PBL and microphysical parameterizations.