Tumor heterogeneity is a consequence of clonal evolution, resulting in a fractal-like architecture with spatially separated main clones, sub-clones and single-cells. As sequencing an entire tumor is not feasible, we ask the question whether there is an optimal clinical sampling strategy that can handle heterogeneity and hypermutations? Here, we tested the effect of sample size, pooling strategy as well as sequencing depth using whole-exome sequencing of ovarian tumor specimens paired with normal blood samples. Our study has an emphasis on clinical application-hence we compared single biopsy, combined local biopsies and combined multi-regional biopsies. Our results show that sequencing from spatially neighboring regions show similar genetic compositions, with few private mutations. Pooling samples from multiple distinct regions of the primary tumor did not increase the overall number of identified mutations but may increase the robustness of detecting clonal mutations. Hypermutating tumors are a special case, since increasing sample size can easily dilute sub-clonal private mutations below detection thresholds. In summary, we compared the effects of sampling strategies (single biopsy, multiple local samples, pooled global sample) on mutation detection by next generation sequencing. In view of the limitations of present tools and technologies, only one sequencing run per sample combined with high coverage (100-300 ×) sequencing is affordable and practical, regardless of the number of samples taken from the same patient.