The U.S. Geological Survey's National Water-Quality Assessment (NAWQA) Program used field matrix spikes (FMSs), field matrix spike replicates (FMSRs), laboratory matrix spikes (LMSs), and laboratory reagent spikes (LRSs), in part, to assess the quality of volatile organic compound (VOC) data from water samples collected and analyzed in more than 50 of the Nation's largest river basins and aquifers (Study Units). The data-quality objectives of the NAWQA Program include estimating the extent to which variability, degradation, and matrix effects, if any, may affect the interpretation of chemical analyses of ground- and surface-water samples. In order to help meet these objectives, a known mass of VOCs was added (spiked) to water samples collected in 25 Study Units. Data within this report include recoveries from 276 ground- and surface-water samples spiked with a 25-microliter syringe with a spike solution containing 85 VOCs to achieve a concentration of 0.5 microgram per liter. Combined recoveries for 85 VOCs from spiked ground- and surface-water samples and reagent water were used to broadly characterize the overall recovery of VOCs. Median recoveries for 149 FMSs, 107 FMSRs, 20 LMSs, and 152 LRSs were 79.9, 83.3, 113.1, and 103.5 percent, respectively. Spike recoveries for 85 VOCs also were calculated individually. With the exception of a few VOCs, the median percent recoveries determined from each spike type for individual VOCs followed the same pattern as for all VOC recoveries combined, that is, listed from least to greatest recovery-FMSs, FMSRs, LRSs, and LMSs. The median recoveries for individual VOCs ranged from 63.7 percent to 101.5 percent in FMSs; 63.1 percent to 101.4 percent in FMSRs; 101.7 percent to 135.0 percent in LMSs; and 91.0 percent to 118.7 percent in LRSs. Additionally, individual VOC recoveries were compared among paired spike types, and these recoveries were used to evaluate potential bias in the method. Variability associated with field spiking, field handling, transport, and analysis was assessed by comparing recoveries between 107 pairs of FMR and FMSR samples. For most VOCs, FMSR recoveries were greater than the paired FMS recoveries. This may result from routinely processing the FMS sample first, allowing a more fluid and efficient technique when processing the FMSR. Degradation was examined by comparing VOC recoveries between 20 pairs of FMS and LMS samples. For all VOCs, the LMS recoveries were greater than FMS recoveries. However, data presented in a previously published VOC stability study were interpreted, and recoveries indicated that VOC degradation should not affect the recovery for most VOCs monitored by the NAWQA Program. Matrix effects were examined by comparing VOC recoveries from 20 pairs of LMS and LRS samples. With the exception of two VOCs, individual recoveries were not significantly different between LMSs and LRSs, indicating that most VOC recoveries are not affected by matrix effects. Additionally, matrix effects should be negligible due to the analytical technique (purge and trap capillary column gas chromatography/mass spectrometry) used for VOC analysis at the U.S. Geological Survey National Water Quality Laboratory (NWQL). The reason for the lower VOC recoveries from FMSs and FMSRs than from LMSs and LRSs may be associated with differences in spiking technique and experience, and to varying environmental conditions at the time of spiking. However, for all spike types, 87 percent of the individual VOC recoveries were within the range of 60 to 140 percent, a range that is considered acceptable by the U.S. Environmental Protection Agency's established analytical method. Additionally, the median recovery for each spike type was within the range of 60 to 140 percent. The excellent VOC recoveries from LMSs and LRSs demonstrate that low VOC concentrations can routinely and accurately be measured by the analytical methods used by the NWQL.