Many challenges still exist in the empirical measurement of population size or density of burrow-nesting procellariiforms. Although reasonable precision of burrow occupancy estimates can be achieved with 10-15 transect (20 entrances per transect) per site, unknown levels of bias in burrow occupancy estimates currently prevents reliable estimation of burrow-nesting procellariiform abundance or harvest rates. Because it is unlikely that biases in burrow occupancy are similar among colonies, valid comparisons among sites may require estimates of absolute abundance rather than relative measures of burrow occupancy. The reliability and precision of matrix models for procellariids will depend primarily on the reliability and precision of adult survival estimates. Sensitivities, elasticities and uncertainties of population growth rate to demographic parameters for models with differing structures and parameterisations showed an overwhelming importance of adult survival in determining population growth rate and results of perturbation analyses. Estimates of adult survival should be a primary focus of any procellariid research program involving assessment of population status, or questions of population response to perturbations. Juvenile survival, pre-breeder survival and emigration rates were also shown to be relatively important in determining population growth rate and perturbation analyses. The sensitivity and elasticity of population growth rate to survival rates for all immature stages combined were similar in magnitude to the sensitivity and elasticity of population growth rate to survival rates for fecund birds. Estimation of survival rates for immature birds should also be given high priority in procellariid research programs. The variability in these parameters among populations needs to be assessed if results are to be generalised beyound specific colonies from which parameters are estimated. There is evidence that selective harvest of heavier Titi chicks occurs on at least some islands. However, analyses of a demographic model incorporating different quality chicks showed even extremely high degrees of selective harvest had little influence on population growth rate or perturbation analyses unless overall harvest levels were very high. Comparison of population growth rate and perturbation analyses of models differing in the level of detail in parameterisation or in the number of stages included in the model, showed negligible differences in results. This suggests that simple models, even if based on only sparse data, are adequate to set research priorities and evaluate population response to perturbations such as for the assessment of conservation management options, evaluation of possible causes of population change and assessment of the effects of harvest.