Background. Trivalent inactivated influenza vaccine (TIV) is reformulated annually to contain representative strains of 2 influenza A subtypes (H1N1 and H3N2) and 1 B lineage (Yamagata or Victoria). We describe a sentinel surveillance approach to link influenza variant detection with component-specific vaccine effectiveness (VE) estimation. Methods. The 2006–2007 TIV included A/NewCaledonia/20/1999(H1N1)—like, A/Wisconsin/67/2005(H3N2)— like, and B/Malaysia/2506/2004(Victoria)—like components. Included participants were individuals ⩾9 years of age who presented within 1 week after influenzalike illness onset to a sentinel physician between November 2006 and April 2007. Influenza was identified by real-time reverse-transcriptase polymerase chain reaction and/or culture. Isolates were characterized by hemagglutination inhibition assay (HI) and HA1 gene sequence. VE was estimated as 1-odds ratio for influenza in vaccinated versus nonvaccinated persons. Results. A total of 841 participants contributed: 69 (8%) were ⩾65 years of age; 166 (20%) received the 2006–2007 TIV. Influenza was detected in 337 subjects (40%), distributed as follows: A/H3N2, 242 (72%); A/H1N1, 55 (16%); and B, 36 (11%). All but 1 of the A/H1N1 isolates were well matched, half of A/H3N2 isolates were strain mismatched, and all B isolates were lineage-level mismatched to vaccine. Age-adjusted estimated VE for A/H1N1, A/H3N2, and B components was 92% (95% CI, 40%–91%), 41% (95% CI, 6%–63%), and 19% (95% CI, –112% to 69%), respectively, with an overall VE estimate of 47% (95%CI,18%–65%). Restriction of the analysis to include only working-age adults resulted in lower VE estimates with wide confidence intervals but similar component-specific trends. Conclusions. Sentinel surveillance provides a broad platform to link new variant detection and the composite of circulating viruses to annual monitoring of component-specific VE.