The Japan Butterfly Conservation Society (JBCS) conducts a project to survey butterflies in gardens to monitor conservation status of butterflies. I have been surveying a butterfly community in a small urban garden for this project. I will present the results of the survey and discuss conservation issues including pollinator–plant relationship in the small urban garden in this paper. Methods: I surveyed in a roadside small green patch (ca. 80 m x 4 m) for 10 years from 2016 in Kyoto city. I recorded butterflies and nectar-sucked plants once a week walking along the periphery road of the green patch for one hour. I used the data that I reported on the garden butterfly monitoring website of JBCS for the analyses. Results: Forty butterfly species visited this garden. The most abundant was licaenid Pseudozizeeria maha followed by papilionid Papilio xuthus, pierid Eurema mandarina, pierid Pieris rapae, nymphalid Argyreus hyperbius and licaenid Lampides boetics in decreasing order. These 6 species occupied 82% of the total individuals counted (6708). These butterflies were all common species and mostly multivoltine. Thirteen species recorded only once or twice were accidental visitors or rare in the urban area. One-third of butterfly species recorded in Kyoto prefecture appeared in this small garden. Species richness of butterflies was stable over 8 years ranging 20 to 27 but it declined in the last two years. The most abundant Pseudozizeeria maha, the second abundant Papilio xuthus and some other species showed increasing trend. Papilio protenor, however, declined. Rising temperature could contribute to the increasing butterfly number but partly. I evaluated richness and abundance of flower plants in the garden using the nectar-sucked plants data. The number of nectar-sucked plant species has been between 30 and 40 in each year. The abundances of most of nectar-sucked plants defined by plant apparency (Feeny 1968) did not chang much during 10 years but two shrubby trees, Lantana and Lespedeza grew lager gradually. Increase in these attractive shrubs might have also contributed to increase in butterfly numbers. I tested a hypothesis that butterflies should be more generalistic in flower selection than plans that should be more specialistic in their pollinators. The niche breadth matrix (Levins 1968) between butterflies and plants indicated that plants tended to be more specialistic than butterflies but not significantly. The butterfly community included generalistic and also specialistic groups.