The effects of light on serum bilirubin subfractions in vitro were investigated by HPLC and four routine methods for bilirubin analysis. By HPLC, the rate of photodegradation of unconjugated bilirubin (Bu) was nearly twice that of monoconjugated bilirubin (mBc) and threefold that of diconjugated bilirubin (dBc); delta bilirubin (Bd) was most stable against photoirradiation. In the diazo method, the rate of photodegradation of direct bilirubin was almost the same as that of the sum of mBc, dBc, and Bd determined by the HPLC method. However, the rate of photodegradation of indirect bilirubin was significantly lower (P < 0.001) than that obtained by HPLC, because approximately 30% of the bilirubin photoproducts reacted with the diazo reagent as indirect bilirubin. The rate of photodegradation of total bilirubin determined by the direct spectrometric method was lower than that determined by the diazo method, but equal to that of the total peak areas of HPLC. In the Ektachem method, bilirubin photoproducts affected total bilirubin negligibly, and Bc and Bu positively, so that the value of Bd decreased. In the bilirubin oxidase method, bilirubin photoproducts were oxidized enzymatically by both the total and direct bilirubin reagents. We re-emphasize the importance of shielding serum from light to avoid generating bilirubin photoproducts that interfere with the accurate determination of serum bilirubin subfractions. We also recommend HPLC analysis as a standard method for bilirubin measurement.