Twenty-seven flavonoids isolated from Dalbergia parviflora with vast structural diversity were screened for inhibitory activity against mushroom and murine tyrosinases using l-DOPA as the substrate. Among the flavonoids tested, only four—khrinone (5), cajanin (9), (3RS)-3′-hydroxy-8-methoxy vestitol (21), and (6aR,11aR)-3,8-dihydroxy-9-methoxy pterocarpan (27)—reacted with mushroom tyrosinase, with IC50 values of 54.0, 67.9, 67.8, and 16.7 μM, respectively, and only compound 27 showed inhibitory activity against murine tyrosinase. With cell-based assays, only compounds 9 and 27 effectively inhibited melanogenesis in B16-F10 melanoma cells (by 34% and 59%, respectively), at a concentration of 15 μM, without being significantly toxic to the cells. However, the crude extract of D. parviflora and some of the flavonoid constituents appeared to increase melanin production in B16-F10 cells, suggesting that there are flavonoids with both inhibitory and stimulatory melanogenesis in the crude extract. Studies on the correlation between the enzyme-based and cell-based assays showed that only the flavonoids with IC50 values below 50 μM against mushroom tyrosinase could inhibit the mammalian tyrosinase, and thus, reduce melanogenesis in B16-F10. Flavonoids with the IC50 values greater than 50 μM, on the other hand, could not inhibit the mammalian tyrosinase, and had either no effect or enhancement of melanogenesis. In conclusion, the tyrosinase enzyme from mushroom is not as selective as the one from mammalian source for the enzyme-based melanogenesis inhibitory screening, and the mammalian cell-based assay appears to be a more reliable model for screening than the enzyme-based one.