This paper is focused on three problematic areas of the index number approach to productivity comparisons: (i) sensitivity of the results that are obtained by applying alternative fixed- and flexible-weight index numbers; (ii) extension of bilateral comparisons to multilateral intertemporal and interspatial comparisons of productivity; (iii) measurement of direct and indirect effects of productivity on costs of production. Direct and indirect effects of relative productivity levels are defined by incorporating the index number approach into the input-output analytical framework. These direct and indirect effects are therefore accounted for by using a modified Leontief inverse matrix. The starting point of the discussion is the observation that, in general, there is no unique way to measure relative productivity levels and that it is important to assess the sensitivity of the results that are obtained by following alternative approaches. Empirical comparative studies show that these results can differ widely.