Abstract The pollination efficiency hypothesis has long been proposed as an explanation for interspecific variation in pollen–ovule ( P : O ) ratios. However, no empirical study on P : O ratios has directly and quantitatively measured pollen transfer efficiency ( PE ). Here, we use a PE index, defined as the proportion of pollen grains removed from anthers that are subsequently deposited on conspecific stigmas, as a direct and quantitative measure of PE . We investigated P : O ratios, pollen removal and pollen deposition in 26 plant species in an alpine meadow, over three consecutive years. Our community survey showed that nearly 5% of removed pollen was successfully deposited on conspecific stigmas. The PE index ranged from 0.01% up to 78.56% among species, and correlated negatively with the P : O ratio across years. This correlation was not changed by controlling for phylogenetic relationships among species, suggesting that the interspecific variation in P : O ratios can be attributed to the probability of pollen grains reaching a stigma. The results indicate that the pollination efficiency hypothesis can help to explain interspecific variation in P : O ratios.