A new replica moulding technique for determining cross-sectional area has been developed. The accuracy of this technique has been investigated. Comparisons have been made with the area micrometer and the laser micrometer to establish the relative accuracy of each technique. The replica technique systematically underestimated the cross-sectional area by 6.2%. However, when this shrinkage artefact was corrected for, the residual error was +/- 1.8%. Relative to the corrected replica method, it was calculated that the area micrometer underestimated cross-sectional area by 16.2 +/- 6.9% (SD) and that the assumption of a convex cross-section would have caused the laser micrometer to overestimate cross-sectional area by an average of 2.3 +/- 1.5% (SD) for tendon, 4.7 +/- 2.5% for 'round' ligament specimens and 19.1 +/- 6.4% for cruciate ligament bundles with obvious concavities in their cross-sections. The moulding method was thought to be particularly useful for soft tissue specimens with irregular cross-sections.