As a preliminary to the revision of the Griffiths scales, 79 examiners throughout Britain tested 447 infants under the age of 2 years during 1978-1982. Compared with the original standardization sample of 1947-1951, the 1980 sample mean general quotient (GQ) was significantly higher, even after adjustments had been made for a slightly skewed social class distribution. Nor did regional differences between samples account for the GQ differences. Unreliable items were not significantly more common among those where the median age at passing had changed most. Various technical difficulties in representing differences between samples were encountered and dealt with. A population's overall results may be given in terms of mean score increase per month, median age advance in weeks or months, e.g. per year, and increase in general developmental quotient. Findings of differences between populations in age at passing individual items can be given as advances in median age, but are better expressed as ratios of the median ages of the two samples in order to deal with the differences in the chronological age for which items are designed. Only then can items be compared for amount of change. In terms of score and mental age advances the second year appears more changed than the first, but in terms of GQ the reverse is true. The locomotor scale mean differs more than other scales from the original norms for both years of infancy, while the personal-social scale differs more in the first year. Taking the most cautious estimate of differences from the data available on both samples, at least 160 items are now passed earlier than in the 1950 sample and the ratio of 1950 to 1980 ages is greater than 110 for 83 of those. Further study of the Griffiths scales items, in relation to one another and to cultural factors, seems feasible on the basis of steps illustrated in this paper.