Effective management and conservation of wildlife populations require reliable estimates of population size, which can be difficult and costly to obtain. We evaluated how precision in estimates of herd size and abundance varies with sample size and strip width using two field surveys and bootstrap resampling of the field data. We also examined precision under distance sampling and evaluated the cost-effectiveness of both survey techniques. Precision in estimates of abundance increased with increasing sample size and varied with strip width independently of sample size. The hazard rate key function was best for five species in two surveys with contrasting visibility conditions. Precision in density was more sensitive to the number of herds sighted than to variation in herd size and effective strip width for distance sampling. Strip counts produced lower abundance estimates but higher precision than distance sampling. We estimated that distance sampling would cost about US$3.1 km−1 of transect. Strip counts deserve serious consideration for surveys of species that occur at high densities and form large, loose agglomerations but distance methods are suitable for species occurring at moderate to low densities in areas where visibility varies substantially. Distance sampling may thus need to be supplemented by strip counts to efficiently estimate densities of rare, abundant and highly clustered multi-species assemblages of African savanna mammals. In small areas, it may often prove necessary to conduct several surveys to obtain adequate sample sizes for distance models.