Lipid droplets (LDs) are highly dynamic organelles that store neutral lipids, primarily triacylglycerols (TAGs), and are found in many cell types. While their primary function is to store excess energy, LDs are also modified in different disease states and during developmental processes. In many cases, not only the presence, but also the composition, of LDs can be equally important. In humans, LD composition has been linked to diseases such as type 2 diabetes; in plants and yeast, LD composition is relevant for engineering these organisms into biological factories in, e.g., algal bioenergy or food oil production.Therefore, lipid analysis of biological LDs with high speed and accuracy in situ is a very active area of research. Here we present an approach for in situ, quantitative TAG analysis using label-free, coherent Raman microscopy to decipher LD TAG composition in different biochemically complex samples. Our method allows direct visualization of inter-LD compositional heterogeneity of physiological quantities -TAG chain length and number of C=C bondswith sub-micrometer spatial resolution within 5-100 milliseconds. Combined with virtually no sample preparation, this approach should enable rapid and accurate TAG LD analysis for a variety of applications.