Microsatellites are highly polymorphic markers which have been used in a wide range of genetic studies. In recent years, various sources of next-generation sequencing data have been used to develop new microsatellite loci, but compared with the more common shotgun genomic sequencing or transcriptome data, the potential utility of RAD-seq data for microsatellite ascertainment is comparatively under-used. In this study, we employed MiddRAD-seq data to develop polymorphic microsatellite loci for the endangered yew species Taxus florinii. Of 8,823,053 clean reads generated for ten individuals of a population, 94,851 (1%) contained microsatellite motifs. These corresponded to 2993 unique loci, of which 526 (18%) exhibited polymorphism. Of which, 237 were suitable for designing microsatellite primer pairs, and 128 loci were randomly selected for PCR validation and microsatellite screening. Out of the 128 primer pairs, 16 loci gave clear, reproducible patterns, and were then screened and characterized in 24 individuals from two populations. The total number of alleles per locus ranged from two to ten (mean=4.875), and within-population expected heterozygosity from zero to 0.789 (mean=0.530), indicating that these microsatellite loci will be useful for population genetics and speciation studies of T.florinii. This study represents one of few examples to mine polymorphic microsatellite loci from ddRAD data.