Although life histories can be modelled with great generality using projection matrices, for organisms with life histories that can be accurately described by a simplified set of parameters, e.g. when adult fecundity and mortality are independent of age, more accurate estimates of life table parameters and of population growth rate and its standard error can be readily obtained. Here an analytic method for calculating approximate confidence intervals for population growth rate is given for two-stage life histories that can be described by four variables representing age at first breeding, fecundity per unit time, and juvenile and adult survivorships per unit time. The method is applied to experimental data on Capitella sp. I obtained by Hansen et al., and quite good agreement is found between the analytic and bootstrap estimates of the standard error of λ. The analytic estimates were a little conservative, probably because of the way the action of mortality was modelled. Alternative life-history models are briefly discussed, and the desirability of formulating life-history models so that the variables involved are independent of each other is stressed. Analytic estimates of λ may be biassed if an inappropriate model is chosen or if variables are not independent and the correlations between them are not measured. To allow for these possibilities, where necessary a conservative approach should be taken to significance testing using the analytic method.