The aim of this paper is to report the use of spectroscopic techniques for the measurement of basic wood density in Eucalyptus nitens, using samples in a state similar to that found in the field; so the current design can later be extended to a portable instrument that could eventually estimate the wood characteristics of a standing tree. Model calibration was carried out using wet samples taken from eight-year-old trees. Each sample was scanned to acquire two types of spectra: Raman and near infrared (NIR). Several pre-processing techniques were applied to the spectra in order to obtain the best possible prediction models for wood basic density using partial least squares (PLS) regression. The model selection criteria were based on maximising the coefficients of determination R-c(2) and r(v)(2) and minimising the root mean square error (RMSE). The potential of NIR for this purpose was demonstrated; r(v)(2) up to 0.87 and RMSEP down to 8.6 kg m(-3) were obtained. Raman spectroscopy proved to be unfit for the application, yielding extremely low correlation values and models with low predictive capabilities.