A 4-year study compared 4 methods of subsampling ‘McIntosh’ apples for mineral analyses intended for use in predicting postharvest senescent breakdown. Concentrations of Ca, Mg, K, P, and N were determined in outer cortex, whole fruit, and juice in 1979 and 1980, and in outer cortex and cortical plugs in 1981 and 1982. Multiple linear regression equations were developed for each sampling in 1979 and 1981 relating the natural logarithms (In) of Ca and P concentrations and Mg:Ca, K:Ca, and N:Ca ratios to In (breakdown percentage +1). Effectiveness of each equation as a predictor of breakdown during and following storage was tested using data collected the following year. Paired t tests comparing predicted breakdown, using 1979 equations and 1980 mineral analyses, to breakdown observed after storage in 1980-1981 showed that outer cortex- and whole fruit-based predictions were not significantly different from observed breakdown, but juice equations predicted significantly more breakdown than was observed. Correlation coefficients relating actual to predicted breakdown were positive and significant for all 3 methods. In 1981, separate equations were developed to predict fruit breakdown after air and after controlled atmosphere (CA) storage. Paired t tests showed the predicted incidence of breakdown following CA storage was not significantly different from actual breakdown incidences for either outer cortex or cortical plug equations. Breakdown incidence following air storage was less than predicted. In all instances, correlation between actual and predicted breakdown incidence was positive and significant. Outer cortex, whole fruit, and cortical plug analyses showed promise as breakdown predictors, but juice analyses were ineffective.