The role of antiretroviral history and genotypic resistance information as predictors of the first treatment interruption (TI) length in a CD4(+) cell count and plasma viremia-guided TI study (GTI) was assessed. Drug-resistance mutations (DRMs) were monitored in chronically HIV-1-infected subjects who underwent GTI. Patients were retrospectively classified into those who received monotherapy or dual therapy prior to HAART (pre-HAART group, n = 44) or directly initiated HAART (HAART group, n = 43). DRMs were assessed by population-based sequencing of proviral DNA at baseline and plasma RNA monthly during TI up to 180 weeks. Univariate and multivariate Cox's proportional hazard models were used to determine time off therapy predictors. The emergence of viruses with DRMs during TI was 5.1-fold more likely in pre-HAART than in HAART patients. The presence of DRMs in proviral DNA or plasma RNA was associated with shorter time off therapy. An accumulation of three or more DRMs duplicated the risk of restarting therapy with respect to having one or two mutations. Regardless of the number of DRMs, the presence of K70R or T215F/Y predicted the shortest TI time. Multivariate analyses adjusted by nadir CD4(+) counts supported the presence of DRMs in plasma HIV-1 RNA, and specifically the K70R or T215F/Y, as potent predictors of time off therapy. A history of monotherapy or dual therapy, accumulation of three or more key DRMs in the HIV-1 polymerase, and/or the presence of substitutions K70R or T215F/Y were associated with shorter time off therapy during GTI. A genotypic profile could provide clinicians with a predictive tool for time off therapy when TI is required in patients with suppressed viremia in whom nadir CD4(+) count is not available.