Green spaces in the city are important for human wellbeing, but are also zones in which humans can become infected with zoonotic diseases. Therefore, there is a need to understand how infection risk is related to green space characteristics, wildlife communities and connectivity with rural areas hosting reservoir populations of hosts. Our hypothesis is that wildlife hosts in urban green spaces, and thereby the prevalence of questing ticks and their Lyme disease causing pathogens (Borrelia burgdorferi s.l.), can be partly predicted based on green space characteristics as well as measures of connectivity to known source areas. We sampled ticks in twenty-two green spaces during Spring (2014 and 2016) and Autumn 2016, located along an urbanization gradient in Antwerp (Belgium). More than 18,000 m2 was sampled, with tick densities ranging from 0 to 386 individuals/100 m2. We estimated connectivity using the least-cost algorithm as either the cost distance to the nearest green space, or to a known population of roe deer (Capreolus capreolus), known to be an important tick propagation host. Both connectivity measures turned out to be correlated, reflecting a gradient in green space isolation from the periphery to the urban center. In 87% of plots where ticks were trapped, at least one Borrelia-infected tick was found. The overall Borrelia-prevalence in nymphs was 17.8%, in adults 32.6%. Density of infected ticks decreased with urbanization and increased with connectivity. Nymphs in larger green spaces were more likely to be infected. While density and infection prevalence for adults increased with the amount of neighboring agricultural land, the larval density and nymphal infection prevalence decreased. Interestingly, the proportion of Borrelia genospecies associated with birds or mammals was comparable in rural and (sub)urban areas (bird/mammal: 0.38), suggesting that even in small green spaces Borrelia infections can persist in local host populations.