STUDY OBJECTIVE: To describe and compare magnitude of class inequalities in women’s health detected with four occupation-based class measures: individual, conventional household (male dominant), gender neutral household, and combined household. DESIGN: Cross sectional study, using health data obtained by physical examination, laboratory analysis, and self report. SETTING: A large pre-paid health maintenance organisation in Oakland, CA (US). PARTICIPANTS: 686 women (90% white) enrolled in Examination II of the Kaiser Permanente Women Twins Study (1989-1990). MAIN RESULTS: The proportion of women categorised as ‘working class’ equalled 45, 30, and 21 per cent, respectively, for the individual level, gender neutral household, and conventional household class measures. Class inequalities in health, comparing women categorised as working class with non-working class, generally were stronger using the gender neutral household class measure, compared with the conventional household or individual class measure; in the case of ‘fair or poor’ health, the respective odds ratios and 95% confidence intervals (adjusted for age and marital status) were 1.9 (1.1, 3.4), 1.5 (0.9, 2.5), and 1.3 (0.8, 2.2), while for high post-load glucose levels, they were 1.7 (1.1, 2.6), 1.2 (0.8, 1.7), and 1.3 (0.9, 1.8). The combined household class measure yielded effect estimates comparable to those of the gender neutral household class measure but with less precision, because of smaller strata. CONCLUSIONS: Epidemiological studies concerning class inequalities in women’s health should use the gender neutral household class measure or, if sample size is sufficiently large, the combined household class measure.