Current methods for determining the maturity of vanilla (Vanilla planifolia Andrews) beans are unreliable. Yellowing at the blossom end, the current index, occurs before beans accumulate maximum glucovanillin concentrations. Beans left on the vine until they turn brown have higher glucovanillin concentrations but may split and have low quality. To find a better index, changes in bean dimensions, dry matter (DM) and glucovanillin accumulation were followed over four seasons. Beans reached their maximum length and width before glucovanillin and DM accumulation began, and there was no clear point when their weight ceased to increase. Beans harvested when the blossom end turned yellow had lower glucovanillin concentrations than beans harvested later; glucovanillin accumulation continued until beans started to turn brown on the vine. Therefore, bean colour, dimensions, weights and glucovanillin content are not useful indicators of maturity, and the lack of visual changes until beans senesce explains why commercially cured beans vary in quality. DM accumulation reached a maximum before browning and was highly correlated with glucovanillin content; the central portion was the most representative of the entire bean. Therefore, optimum harvest time occurs when DM accumulation slows and should be measured in the central portion of beans. Two near infrared spectrometers using interactance geometry were trialled for non-invasive assessment of DM. Cross validation r and RMSECV values of 0.87 and 1.76, respectively, for a unit using wavelengths between 1100 and 2300nm, and 0.82 and 1.05 for a portable unit using wavelengths between 800 and 1050nm were obtained from the second derivative of absorbance spectra. The latter unit allows infield monitoring of maturation.