When monofilament nylon strings of different tensions are used on a plucked instrument like the guitar or lute, it is readily observed that a lower-tension string tends to give a brighter sound. The elementary linear theory of a plucked string predicts no such effect. Linear theory can be extended to include bending stiffness and frequency-dependent damping: detailed measurements on a range of nylon strings are analysed to enable this extended model to be made quantitatively accurate. It is shown that a small influence of the string gauge is predicted, but synthesis based on this model does not match measurements or perceived differences of sound quality. Two nonlinear effects are then explored. The familiar effect of coupling between transverse and axial vibration in the string is shown to account for some observed features, but still does not explain the large difference of sound. The evidence from measurements strongly suggests that the mechanism responsible for the sound differences involves nonlinear interaction between the string and other parts of the structure, such as the frets. Because a lower-tension string tends to be played at larger amplitude, such “buzzing” interactions are more likely to occur. Implications for makers and players of instruments like the lute are discussed.