The decline of kereru (Hemiphaga novaeseelandiae) may limit dispersal of large-seeded plants in New Zealand, but the consequences of this are unknown. I determined kereru disperser effectiveness by modelling seed dispersal distances (using seed retention times and movement patterns). Mean seed retention time was significantly longer for larger-seeded species, ranging from 37-181 minutes. Wild radiotracked kereru were sedentary, remaining at one location for up to 5.25 hours. The mean flight distance was 77 m and the maximum was 1, 457 m. Estimated mean seed dispersal distances for tawa (Beilschmiedia tawa), puriri (Vitex lucens), and fivefinger (Pseudopanax arboreus) were 95, 98, and 61 m respectively. Kereru dispersed 66-87% of ingested seeds away from the parent tree, with 79-88% of seeds dispersed <100 m and < 1% dispersed over 1,000 m. In a field seed-fate experiment, "pre-human" conditions (cleaned seeds, low density, away from parent, and protected from mammals) increased survival compared to "post-human" conditions (whole fruits, high density, under parent, not protected) for both taraire (Beilschmiedia tarairi; 15% vs. 2% survival to one year respectively) and karaka (Corynocarpus laevigatus; 60% vs. 11% to two years, respectively). Fruit diameter varied considerably within karaka, taraire, and tawa, although theoretically not enough for them to be swallowed by other birds. Nevertheless, other birds are reported to occasionally take fruits of nearly all large-seeded species. Small tawa seeds produced smaller seedlings in the glasshouse; therefore selection of only smaller seeds by alternative dispersers may negatively affect tawa recruitment. Kereru are generally not gape-limited, and fruit size preferences were independent of mean fruit size. Kereru provide effective dispersal by moving most seeds away from the parent, and enhancing seed and seedling survival. Therefore, both dispersal failure and introduced mammals negatively affect the regeneration of large-seeded trees in New Zealand.