Knowledge of intrinsic tumor heterogeneity is vital for understanding of tumor progression mechanisms as well as for providing efficient treatments. In situ proteomic profiling of tumors is a powerful technology with potential to enhance our understanding of tumor biology, but sources of variability due to patient and tumor heterogeneity are poorly understood and are the topic of this investigation. Clarification of variability within case and between cases is also important for designing future studies. Direct protein profiling by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) is a sensitive and powerful technology for obtaining hundreds of protein expression peaks from a thin tissue section. By combining robotic microspotting and laser capture microdissection with MALDI MS, we acquired multiple spectra per case to evaluate inter- and intra-case variability in human colorectal tumor and murine cecal carcinoma. We detected 256 peaks from 164 samples of 111 patients, which consisted of 55 normal colorectal mucosal samples, 24 adenomas, 71 primary carcinomas, and 14 hepatic metastases. In addition, we detected 291 peptide/protein peaks from 34 orthotopically transplanted murine cecal carcinomas and 14 hepatic metastases. In human colorectal samples, we observed that proteomic profiling in adenomas was more homogeneous across patients than in normal mucosa specimens (p=0.0008), but primary carcinoma exhibited greater heterogeneity than normal mucosa and adenomas (both p<0.0001). Murine cecal carcinomas were homogeneous within and between carcinomas, while their hepatic metastases tended toward greater intra-tumor differences (p<0.0001). Inter- and intra-case variability was approximately equal for many protein peaks. Acquiring up to 5 subsamples per case could reduce the total number of cases required, but further reduction from additional subsampling was modest unless intra-case variability comprises a greater proportion of total variation (e.g. >70%). In summary, this study characterizes intra- and inter-case variability of high-throughput protein expression in colorectal tumors, and provides guidance for the sample numbers required for in situ proteomic studies.