A stage- and age-structured model was constructed to simulate stage-abundance patterns of Rhincalanus gigas in a data set consisting of over 80 yr of net-catch observations in the Scotia Sea. The model was initialised with the observed annual abundances of the Cl stage and the population developed according to pre-defined developmental stage durations, which varied according to life-cycle phenotype. Better fits to net-catch observations were achieved by models that allowed a number of different life-cycle phenotypes to co-exist in the population. In particular, a model in which 71 % of individuals reached adulthood in 2 yr (2 yr phenotype) and 29 % in 1 yr (1 yr phenotype) achieved the best fit. Of the 2 yr phenotypes, most individuals spent their 1st winter as a CIII, although a fraction passed this period as a CIV or CV. The 1 yr phenotypes entered their 1st winter as a CV but moulted through to adulthood before the following spring. During the productive period, the mortality rate of the early developmental stages was 0.1 d(-1), but this fell to 0.007 d(-1) as individuals developed beyond stage CIII. During the winter, the mortality rate fell further to 0.003 d(-1). Such rates meant that around 1.5 % of the copepodite population lived for 3 yr or more. Many of these spent 2 yr as an adult. Quantitative descriptions of development and mortality rates in the later stages of long-lived copepods are relatively few because of the difficulty in distinguishing the many generations in a typical population. As well as being the first to determine these rates in R. gigas, this study provides a methodological framework for determining such rates in other copepods with multi-year life cycles.