Elevated plasma homocysteine has been identified as a strong and independent risk factor for cardiovascular diseases, and recently, it has been associated with the development of dementia in older adults. Selected ion-monitoring isotope-dilution LC/MS (electrospray) has been developed and evaluated as a reference method for the accurate determination of total homocysteine in human plasma. Homocysteine is quantitatively isolated from plasma via the use of anion-exchange resins and then detected and quantified in stabilized plasma extracts with selected ion-monitoring LC/MS. This method is shown to be highly comparable to LC/MS/MS determinations in terms of its analytical accuracy and precision, yet this alternative measurement approach does not necessitate the enhanced instrumentation or added expense required of tandem MS/MS determinations. LC/MS detection of homocysteine was linear (standard error of the estimate for the regression line was 0.0323) over 3 orders of magnitude, and the calculated limits of detection and quantification were 0.06 micromol/L (0.12 ng on column) and 0.6 micromol/L (1.2 ng on column), respectively. Independent calibration curves showed excellent linearity (r2 > or = 0.996) between 0 and 25 micromol/L homocysteine over a 3-day period. The accuracy and precision of total homocysteine measurements for patient samples and quality control pools using LC/MS were compared to total homocysteine measurements using LC/MS/MS, GC/MS, FPIA, and LC-FD. LC/MS performed well in relation to the other homocysteine methods in terms of its capability to accurately quantify plasma homocysteine over the normal range (5-15 micromol/L).