In radiotracer studies, the estimation of turnover time usually depends on the assumption of steady-state compartmental precursor-product (SCP) behaviour in the pools being studied. Deviation from SCP behaviour is currently measured from the extent of hysteresis, R, in an 'area plot': a plot of the time-integrated difference between precursor and product specific activities (spec. act.) against product spec. act. We propose two approaches to evaluating the statistical significance of apparent deviations from SCP behaviour in experimental data. The first constructs the sampling distribution of R. The second compares the variance in replicate data with the value of the objective function minimized in the least squares estimation of product turnover time. We show that this test closely approximates a variance ratio (F) test. Applied to [methyl-3H]choline and [14C]acetate tracer data from lung phospholipid pools in eupneic and hyperpneic rats, the analysis rejects SCP relationships between lamellar bodies and two subfractions of alveolar lavage material.