Systematic variations in methods for the storage, digestion and measurement of lake water samples for phosphorus determinations show that these methods are very robust. Frozen samples can be kept for at least one year without any detectable change in total phosphorus, [TP], but the concentrations of P subfractions may change. Variations in the amount of oxidant (0.05 to 0.8 g potassium persulphate/40 ml sample), in the duration (10–90 min), or pressure (100 vs 200 k Pa) of oxidation, and in the amount of sample lost in autoclaving had no effect on estimated [TP]. The exact composition of acid molybdate “mixed reagent” also proved not to be critical, for when each of the components was varied in turn, similar results were obtained with 5 to 10 g ammonium molybdate, 2.5 to 20 g ascorbic acid, 0.025 to 0.4 g potassium antimony tartrate and 55 to 100 ml sulphuric acid per liter of mixed reagent. No significant difference was obtained by varying the amount of mixed reagent from 1 to 7.5 ml/40 ml sample, so some economy may be achieved by reductions in the amount of reagent used. Colour development was stable for up to 8 hours after reagent addition. Refrigerated mixed reagent was stable for up to 48 h, and ascorbic acid solution for 72 h, although storage appeared to increase variation. The insensitivity of this analysis to substantial variations no doubt contributes greatly to the method's success.