The study examined the effectiveness of a tutoring program on counting and number skills for trainable mentally retarded (TMR) and educable mentally retarded (EMR) students (5-14 years old). Experimental Ss received individualized instruction based on counting games while control Ss received instruction on objectives not related to counting. Analysis is presented of pre- and posttests on oral counting, counting transfer, counting by 10, enumeration and production of objects, enumeration transfer, production transfer, cardinality rule, subitizing, finger presentation of 1 to 10, order-irrelevance principle, and equivalence. Pretesting data suggested that basic counting skills cannot be taken for granted in retarded populations. The training was reasonably successful in extending Ss' oral counting sequence, suggesting that short-term intensive individual tutoring that focuses on count patterns is useful, perhaps especially with TMR pupils. Training was not generally successful in producing transfer. Findings suggested that if the S's cooperation can be obtained, oral counting training can be effective with mentally retarded children with relatively low mental ages. (CL)