Binding interactions often involve heterogeneous samples displaying a distribution of binding sites that vary in affinity and binding enthalpy. Examples include biological samples like proteins and chemically produced samples like modified cyclodextrins. Experimental studies often ignore sample heterogeneity and treat the system as an interaction of two homogeneous species, i.e. a chemically well-defined ligand binding to one type of site. The present study explores, by simulations and experiments, the impact of heterogeneity in isothermal titration calorimetry (ITC) setups where one of the binding components is heterogeneous. It is found that the standard single-site model, based on the assumption of two homogeneous binding components, provides excellent fits to simulated ITC data when the binding free energy is normally distributed and all sites have similar binding enthalpies. In such cases, heterogeneity can easily go undetected but leads to underestimated binding constants. Heterogeneity in the binding enthalpy is a bigger problem and may result in enthalpograms of increased complexity that are likely to be misinterpreted as two-site binding or other complex binding models. Finally, it is shown that heterogeneity can account for previously observed experimental anomalies. All simulations are accessible in Google Colab for readers to experiment with the simulation parameters.