Objective: Contacts of pulmonary tuberculosis (TB) cases are at high risk of TB infection and progression to disease. Close and household contacts and those <5 years old have the highest risk. Isoniazid preventive therapy (IPT) can largely prevent TB disease among infected individuals. International and Peruvian recommendations include TB contact investigation and IPT prescription to eligible contacts. We conducted a study in Lima, Peru, to determine the number of close and household contacts who were evaluated, started on IPT, and who completed it, and the factors associated to compliance with national guidelines. Methods: We conducted a longitudinal retrospective study including all TB cases diagnosed between January 2015 and July 2016 in 13 health facilities in south Lima. Treatment cards, TB registers and clinical files were reviewed and data on index cases (sex, age, smear status, TB treatment outcome), contact investigation (sex, age, kinship to the index case, evaluations at month 0, 2 and 6) and health facility (number of TB cases notified per year, proportion of TB cases with treatment success) were extracted. We tabulated frequencies of contact evaluation by contact and index case characteristics. To investigate determinants of IPT initiation and completion, we used generalised linear mixed models. Results: A total of 2323 contacts were reported by 662 index cases; the median number of contacts per case was four (IQR, 2–5). Evaluation at month 0 was completed by 99.2% (255/257) of contacts <5 and 98.1% (558/569) of contacts aged 5–19 years. Of 191 eligible contacts <5 years old, 70.2% (134) started IPT and 31.4% (42) completed it. Of 395 contacts 5–19 years old, 36.7% (145) started IPT and 32.4% (47) completed it. Factors associated to not starting IPT among contacts <5 years old were being a second-degree relative to the index case (OR 6.6 95CI% 2.6–16.5), not having received a tuberculin skin test (TST) (OR 3.9 95%CI 1.4–10.8), being contact of a smear-negative index case (OR 5.5 95%CI 2.0–15.1) and attending a low-caseload health facility (OR 2.8 95%CI 1.3–6.2). Factors associated to not starting IPT among 5–19 year-olds were age (OR 13.7 95%CI 5.9–32.0 for 16-19 vs. 5–7 years old), being a second-degree relative (OR 3.0 95%CI 1.6–5.6), not having received a TST (OR 5.4, 95%CI 2.5–11.8), being contact of a male index case (OR 2.1 95CI% 1.2–3.5), with smear-negative TB (OR 1.9 95%CI 1.0–3.6), and attending a high-caseload health facility (OR 2.1 95%CI 1.2–3.6). Factors associated to not completing IPT, among contacts who started, were not having received a TST (OR 3.4 95%CI 1.5–7.9 for <5 year-olds, and OR 4.3 95%CI 1.7–10.8 for those 5–19 years old), being contact of an index case with TB treatment outcome other than success (OR 9.3 95%CI 2.6–33.8 for <5 year-olds and OR 15.3 95%CI 1.9–125.8 for those 5–19 years old), and, only for those 5–19 years old, attending a health facility with high caseload (OR 3.2 95%CI 1.4–7.7) and a health facility with low proportion of TB cases with treatment success (OR 4.4 95%CI 1.9–10.2). Conclusions: We found partial compliance to TB contact investigation, and identified contact, index case and health facility-related factors associated to IPT start and completion that can guide the TB programme in increasing coverage and quality of this fundamental activity.