Three-dimensional (3D) cell culture is a rapidly emerging field, which mimics some of the physiological conditions of human tissues. In cancer biology, it is considered a useful tool in predicting in vivo chemotherapy responses, compared with conventional two-dimensional (2D) cell culture. We have developed a novel 3D cell culture model of osteosarcoma composed of aggregated proliferative tumour spheroids, which shows regions of tumour heterogeneity formed by aggregated spheroids of polyclonal tumour cells. Aggregated spheroids show local necrotic and apoptotic regions and have sizes suitable for the study of spatial distribution of metabolites by mass spectrometry imaging (MSI). We have used this model to perform a proof-of-principle study showing a heterogeneous distribution of endogenous metabolites that colocalise with the necrotic core and apoptotic regions in this model. Cytotoxic chemotherapy (doxorubicin) responses were significantly attenuated in our 3D cell culture model compared with those of standard cell culture, as determined by resazurin assay, despite sufficient doxorubicin diffusion demonstrated by localisation throughout the 3D constructs. Finally, changes to the distribution of endogenous metabolites in response to doxorubicin were readily detected by MSI. Principal component analysis identified 50 metabolites which differed most in their abundance between treatment groups, and of these, 10 were identified by both in-software t test and mixed-effects analysis of variance (ANOVA). Subsequent independent MSIs of identified species were consistent with principle component analysis findings. This proof-of-principle study shows for the first time that chemotherapy-induced changes in metabolite abundance and distribution may be determined in 3D cell culture by MSI, highlighting this method as a potentially useful tool in the elucidation of chemotherapy responses as an alternative to in vivo testing.