This article adopts a non-parametric approach to explore the relation between size, capital intensity and productivity in a set of Portuguese manufacturing sectors. The article makes use of 2007 data from firm’s balance sheets and income statements in sectors “food and beverages”, “clothing”, “manufactured non-metallic mineral products” and “metallic products, except machinery and equipment”. In 2007, these four sectors represented almost half of the total number of manufacturing firms, more than one third of gross value added and sales and more than forty per cent of employment and capital stock in the Portuguese manufacturing sector. Firstly, the article presents basic descriptive statistics regarding the distribution of firms along the selected variables. Secondly, the analysis is enlarged by estimating robust conditional kernel distributions for the pairs of variables capital intensity-productivity, size-productivity and size-capital intensity. The unconditional distributions for the selected variables reveal some similarities between sectors. There is substantial heterogeneity within sectors but firms are concentrated in classes that correspond to small size, low capital-labour ratios and small number of workers. The conditional distributions reveal that the largest firms in terms of sales tend to be those with higher capital-labour ratios and these two characteristics tend to lead to higher levels of apparent labour productivity.