Sensory peripheral neuropathy (PN) remains a common complication in HIV-positive patients despite effective combination anti-retroviral therapy (ART). Data on PN on second-line ART is scarce. We assessed PN using a standard tool in patients failing first-line ART and for 96 weeks following a switch to PI-based second-line ART in a large Randomised Clinical Trial in Sub-Saharan Africa. Factors associated with PN were investigated using logistic regression. Symptomatic PN (SPN) prevalence was 22% at entry (N = 1,251) and was associated (p < 0.05) with older age (OR = 1.04 per year), female gender (OR = 1.64), Tuberculosis (TB; OR = 1.86), smoking (OR = 1.60), higher plasma creatinine (OR = 1.09 per 0.1 mg/dl increase), CD4 count (OR = 0.83 per doubling) and not consuming alcohol (OR = 0.55). SPN prevalence decreased to 17% by week 96 (p = 0.0002) following similar trends in all study groups (p = 0.30). Asymptomatic PN (APN) increased over the same period from 21 to 29% (p = 0.0002). Signs suggestive of PN (regardless of symptoms) returned to baseline levels by week 96. At weeks 48 and 96, after adjusting for time-updated associations above and baseline CD4 count and viral load, SPN was strongly associated with TB (p < 0.0001). In summary, SPN prevalence was significantly reduced with PI-based second-line therapy across all treatment groups, but we did not find any advantage to the NRTI-free regimens. The increase of APN and stability of PN-signs regardless of symptoms suggest an underlying trend of neuropathy progression that may be masked by reduction of symptoms accompanying general health improvement induced by second-line ART. SPN was strongly associated with isoniazid given for TB treatment.