Better and steady fruit quality evaluation at harvest is a major challenge for commercial growers of apples in Denmark. Those fruits not meeting the requirements for the fresh market traditionally go to the juice concentrate industry where low cost products are obtained. Special fruit qualities are needed to develop commodities that can obtain a premium added value on the market. Nowadays in the food industry, quality evaluation is commonly performed non-destructively by means of optical sensors such as spectrometers, hyperspectral and multispectral cameras, that allow rapid measurements of fruit composition. These methods are desirable to replace the traditional destructive ones in terms of optimization of resources and sampling. In preliminary experiments, we tested visible/NIR spectroscopy for its ability to predict firmness, which is a relevant fruit quality trait connected to juice making as well as for eating apples. Invasive and non-invasive measurements of firmness, on the shaded and exposed side of the fruits were carried out for three Danish apple cultivars of known commercial usage. Resulting data determined wavelengths between 415 to 715 nm to be predictive for firmness. A PLS model for all three cultivars combined using spectral reflectance data in this range had r=0.79 and RSME=0.83 N/cm2.