Abstract Compiling syllable type inventories from children’s spontaneous speech is all but straightforward: so far studies vary considerably in their methodologies and consequently the selection of the speech samples differs. This paper shows that different methodologies for selecting a speech sample lead to substantial differences in syllable inventories. Two main sources for this variation are explored: differences in size and content of the speech samples. Both factors influence the results significantly and this questions the comparability of previous study results. An empirical procedure to investigate syllable type development is proposed to overcome such methodological problems, and this procedure is implemented to provide an initial empirically sound assessment of the acquisition of syllable types in Dutch speaking toddlers.