This report updates a 1981 study of mortality at three United Kingdom chromate producing factories, and pays special attention to workers engaged after major plant and process changes were completed during 1958-60. The study covers 2298 payroll workers in post on 1 January 1950 or entering employment up to 30 June 1976 and remaining at least one year, with mortality observed up to 31 December 1988. Expected numbers of deaths were calculated from national death rates adjusted for social class and area differences. At the two largest factories 1422 men starting work before the process changes showed a highly significant excess of deaths from lung cancer (observed deaths/adjusted expected deaths (obs/expA) 175/88.97, adjusted standardised mortality ratio (SMRA) 197). They also had a significant excess of deaths from nasal cancer (obs/expA 4/0.26, SMRA 1538); the four affected men all had over 20 years of employment. Six hundred and seventy seven men starting work after the completion of process changes showed no excess of lung cancer deaths (obs/expA 14/13.7, SMRA 102, 95% confidence interval (95% CI) 56-171), but the possibility of the risk persisting at a reduced level cannot be excluded. The risk among earlier entrants affected men with two or more years of employment and was highest among those working for 10 years or longer (SMRA 225). The relative risk was already raised 5-14 years after first employment; it was highest in the 25-40 year period, but was still raised 50 years after first exposure. The risk showed most clearly at young ages, with a SMRA of 355 for ages under 50 (obs/expA 21/5.91). The excess was greater among men in jobs with high exposure to chromates (obs/expA 151/61.73, SMRA 245) than among men in less exposed jobs (obs/expA 21/19.57, SMRA 107). Less evidence of a lung cancer excess was found among 199 men employed at a third small factory (obs/expA 12/9.91, SMRA 121). No risk was apparent in further groups of 214 salaried works staff (obs/expA 1/2.53), or 95 workers at an adjacent fertiliser plant (obs/expA 3/3.95). No significant excesses of deaths from cancers of other sites were found.