A non-destructive indicator of seed water content could significantly help crop scientists with assessment of the effects of environmental conditions during drying on grain qualities or on seed physiological quality. This is particularly important for grain legumes which simultaneously bear pods of different ages. Visual assessment of pod colour has so far been used to date grain legume stages, but now colour can be easily and accurately measured with a portable spectrophotometer. Relationships between the spectrophotometer measurements and the pod and seed water contents were tested in various climatic contexts (3 years: 2000, 2001, 2002; field or greenhouse, two or three sowing dates) for two bean cultivars (Booster and Calypso) and also for one pea cultivar (Baccara) in 2003 near Angers, France. Among the different spectrophotometer measurements, hue angle (h) clearly shows the transition from green (h=180 °) to yellow (h=90 °) and then to red (h=0 °). In each context, h and seed water content (SWC) relationships showed the same pattern of three linear phases: first a steady state; then a sharp decrease from green (h=106–108 °) to yellow (h=85–93 °) just before the end of the seed filling stage for Booster or between the end of the seed filling phase and the beginning of seed drying for Calypso and pea; finally, a slow decrease from yellow to ochre (h=75–78 °) during seed drying. For each bean cultivar, the parameters of the linear relationships showed no differences between maturation conditions. Therefore, 6 h classes matching six SWC classes could be defined over a wide range of SWC between 0·56 and 0·2 g/g for Booster. However for Calypso and pea, only 3 h classes could be defined because of the tight relationships between h and SWC during the end of seed drying, which can be explained by pod walls drying faster than seeds. Hence, spectrophotometer measurements, if calibrated for a given cultivar of a species, could now be used to select pods with seeds of the same water content and therefore to study environmental effects on quality criteria either in controlled conditions or in the field.