The lung cancer mortality experienced by a cohort of 3025 workers from a nickel cadmium battery factory during the period 1946-84 has been investigated. Occupational histories were described in terms of 75 jobs: eight with "high," 14 with "moderate" or slight, and 53 with minimal or zero exposure to cadmium oxide (hydroxide) dust. The Mantel-Haenszel technique applied to prospective (or historical prospective) studies was used to compare the estimated cadmium exposures (durations of exposed employment) of those dying from lung cancer with those of matching survivors in the same year of follow up, while controlling for sex and year, and age of starting employment. Among workers first employed in the period 1923-46, there was some evidence of an association between the risk of dying from lung cancer and duration of employment in "high or moderate" exposure jobs, although the evidence relied heavily on the findings for the single highest exposure category. Among workers first employed in the period 1947-75, there was no evidence whatsoever of such an association.