Previous empirical literature - mainly cross-sectional - has tested the demand-pull hypothesis and found that overall, evidence does not conflict with the idea that innovation may be driven by output. Using a balanced panel of 216 Italian manufacturing firms over the 1995-2000 period, and checking for fixed effects, time, sectoral and size dummies and for the path-dependent nature of R&D, we also find a (barely significant) role of sales in inducing R&D expenditures. However, at the micro level, the demand-pull effect plays a varying role for the different sub-samples of firms. In particular, exporting firms, those which are liquidity-constrained, those not receiving public subsidies and those not heading a business group, seem to be particularly sensitive to sales in deciding their R&D expenditures. These microeconometric results have been obtained using a Least Squares Dummy Variable Corrected (LSDVC) estimator, a recently-proposed panel data technique particularly suitable for small samples.