Meta-analyses are presented of sex differences in (1) the (mental) arithmetic subtest of the Wechsler intelligence tests for children and adolescents (the WISC and WPPSI tests), showing that boys obtained a mean advantage of 0.11d; (2) the (mental) arithmetic subtest of the Wechsler intelligence tests for adults (the WAIS tests) showing a mean male advantage of 0.47d; (3) the digit span subtest of the Wechsler intelligence tests for children and adolescents (the WISC and WPPSI tests), showing that girls obtained a mean advantage of 0.134d; (4) the digit span subtest of the Wechsler intelligence tests for adults (the WAIS tests) showing a male advantage of 0.116d among adults. These results show that the sex differences on mental arithmetic are not consistent with the sex differences on digit span. It is proposed that the reason for this is that mental arithmetic is a measure of working memory capacity while digit span is a measure of immediate memory capacity. If this is accepted, the results indicate that there is virtually no sex difference in immediate memory capacity (measured by digit span) but a small male advantage among children and a substantial male advantage among adults in working memory capacity (measured by mental arithmetic). The results are further interpreted in terms of Kyllonen's theory that working memory capacity is "g." If this is accepted, it follows that males have an advantage in "g" and that the higher average means obtained by men in IQ tests like the WAIS and the Progressive Matrices is attributable to their advantage in "g."