Skeletal muscle fibre architecture provides important insights into performance of vertebrate locomotor and feeding behaviours. Chemical digestion and in situ sectioning of muscle bellies along their lengths to expose fibres, fibre orientation and intramuscular tendon, are two classical methods for estimating architectural variables such as fibre length (Lf) and physiological cross-sectional area (PCSA). It has recently been proposed that Lf estimates are systematically shorter and hence less accurate using in situ sectioning. Here we addressed this hypothesis by comparing Lf estimates between the two methods for the superficial masseter and temporalis muscles in a sample of strepsirrhine and platyrrhine primates. Means or single-specimen Lf estimates using chemical digestion were greater in 17/32 comparisons (53.13%), indicating the probability of achieving longer fibres using chemical digestion is no greater than chance in these taxonomic samples. We further explored the impact of sampling on scaling of Lf and PCSA in platyrrhines applying a bootstrapping approach. We found that sampling-both numbers of individuals within species and representation of species across the clade significantly influence scaling results of Lf and PCSA in platyrrhines. We show that intraspecific and clade sampling strategies can account for differences between previously published platyrrhine scaling studies. We suggest that differences in these two methodological approaches to assessing muscle architecture are relatively less consequential when estimating Lf and PCSA for comparative studies, whereas achieving more reliable estimates within species through larger samples and representation of the full clade space are important considerations in comparative studies of fibre architecture and scaling.