The analysis of ecological networks has gained a very prominent foothold in ecology over the last years. While many publications try to elucidate patterns about the networks, others are primarily concerned with the role of specific species in the network. The core challenge here is to tell specialists from generalists. While field data and observations can be used to directly assess specialisation levels, the indirect way through networks is burdened with problems. Here, I review eight measures to quantify specialisation in pollination networks (degree, node specialisation, betweenness, closeness, strength, pollination support, Shannon's H and discrimination d'), the first four being based on binary, the others on weighted network data. All data and R-code are available as supplement and can be applied beyond pollination networks. The indices convey different concepts of specialisation and hence quantify different aspects. Still, there is some redundancy, with node specialisation and closeness quantifying the same properties, as do degree, betweenness and Shannon's H. Using artificial and real network data, I illustrate the interpretation of the different indices and the importance of using a null model to correct for expectations given the different observed frequencies of interactions. For a well-described network the distributions of specialisation values do not differ from null model expectations for most indices. Finally, I investigate the effect of cattle grazing on the specialisation of an important pollinator in eight replicated pollination networks as an illustration of how to employ the specialisation indices, null models and permutation-based statistics in the analysis of specialisation in pollination networks.