The n-alkane method was developed in temperate areas as a tool to estimate voluntary intake (VI) at pasture. The present study aimed to investigate the performance of n-alkanes as markers for estimating VI of steers (mean live weight 213 kg) offered a range of tropical grass hays and lucerne. Tropical and temperate forages have different n-alkane profiles and little is known about the issues which affect the accuracy of the method under tropical conditions. In two pen experiments (no. = 20 and no. = 24) n-alkanes were dosed using intraruminal controlled-release devices. Actual mean voluntary dry matter intakes for the diets ranged from 3·12 to 4·60 kg/day and actual mean dry-matter digestibility varied between 439 and 620 g/kg. n-Alkane profiles (C30 to C36) of the diets and the faeces for each animal were determined using gas chromatography. The recovery of each n-alkane was determined for each animal. Recoveries of n-alkanes were highly variable and generally varied between diets and between experiments. When adjacent n-alkanes were used to estimate VI (ratio method), agreement with actual VI was often poor. Despite this, where the recoveries of n-alkane pairs were similar, group mean VI were accurately estimated. From these data, it is concluded that estimation of VI in cattle offered tropical grass hays or lucerne hay, requires measured recoveries of both dosed and natural plant n-alkanes. The dosed and natural n-alkane pairs having the most similar recoveries should be used in the ratio method to estimate VI.