In studies to determine the inheritance of response-time traits, such as time to seed germination, some viable individuals may fail to respond during an experiment. If these right-censored observations are ignored, sample means and variances will be underestimated. This is illustrated using data from time to seed germination at 9°C for Lycopersicon esculentum (Mill.) fast germinating PI 120256, slow-germinating T3 and their reciprocal F1, F2 and backcross progeny. This paper presents methods to detect and to accommodate right-censored data in generation means analysis. Genetic interpretations derived from corrected and uncorrected estimates of generation means and variances are compared. Correction for right-censoring increased estimates of environmental and phenotypic variances, and decreased heritability estimates. © 1990 Kluwer Academic Publishers.