Information was obtained on the structure of the herds and flocks of a sample of 41 households on the three Maasai Group Ranches in Kenya. The purpose of the survey was to contribute to an intensive study of the livestock production system, and also to assess the amount of information that could be obtained from a herd structure study on its own. Differences in herd structure were related to the livestock wealth of the household, the stage of development of the Group Ranch as well as to the variation in climate, and production alternatives. Data from more than 5100 cattle revealed that Maasai herds contained less steers in 1982 (18%) than in 1968 (22%). The proportion of mature steers had also dropped from 5% (1968) to 2% (1982), with maximum valves of 3·2% on the most developed ranch and 2% for the wealthiest stratum of households. A high percentage of females had been retained (56%) giving a herd structure characteristic of subsistence production with milk as a primary output. This structure also caters for a strong market in immature animals and gives the herd a high potential for recovery after a drought. There appeared to be a trend towards increased use of introduced bulls, notably Sahiwal on the Small East African Zebu, with increased Group Ranch Development. However, this trend was confounded by a decreasing aridity gradient with increased development. On two of the ranches for which 1968 counts were available, cattle numbers were 119% of their previous value whereas small stock were 470%. The reason appears to be that increased sedentarisation without adequate range management has resulted in a severe reduction in grasses, and their replacement by forbs. These plants are more palatable to small stock which are better able to exploit the degraded habitat. Information from more than 2730 sheep and 2300 goats revealed that poor households preferred goats and richer households, sheep. The Small East African Goat was ubiquitous, but the most developed ranch had crossed Dorper onto Maasai sheep whereas the least developed and most arid ranch had introduced Somali Blackhead sheep. The Dorper crossbred catered for a market for mutton whereas the Blackhead was primarily for home consumption of fat. The conclusion is that Group Ranch Development has not projected the Maasai into commercial beef ranching as originally planned, but their production system has not stood still. A study of herd structures was a good way of demonstrating the evolution of the system. The method was simple but was dependent for its success on fairly involved and time-consuming sampling and data collection procedures.