Background: Awareness is growing that cancer can be treated during pregnancy, but the effect of this change on maternal and neonatal outcomes is unknown. The International Network on Cancer, Infertility and Pregnancy (INCIP) registers the incidence and maternal, obstetric, oncological, and neonatal outcomes of cancer occurring during pregnancy. We aimed to describe the oncological management and obstetric and neonatal outcomes of patients registered in INCIP and treated in the past 20 years, and assess associations between cancer type or treatment modality and obstetric and neonatal outcomes. Methods: This descriptive cohort study included pregnant patients with cancer registered from all 37 centres (from 16 countries) participating in the INCIP registry. Oncological, obstetric, and neonatal outcome data of consecutive patients diagnosed with primary invasive cancer during pregnancy between Jan 1, 1996, and Nov 1, 2016, were retrospectively and prospectively collected. We analysed changes over time in categorical patient characteristics, outcomes, and treatment methods with log-binomial regression. We used multiple logistic regression to analyse preterm, prelabour rupture of membranes (PPROM) or preterm contractions, small for gestational age, and admission to the neonatal intensive care unit (NICU). The INCIP registry study is registered with ClinicalTrials.gov, number NCT00330447, and is ongoing. Findings: 1170 patients were included in the analysis and 779 (67%) received treatment during pregnancy. Breast cancer was the most common malignant disease (462 [39%]). Every 5 years, the likelihood of receiving treatment during pregnancy increased (relative risk [RR] 1·10, 95% CI 1·05–1·15), mainly related to an increase of chemotherapeutic treatment (1·31, 1·20–1·43). Overall, 955 (88%) of 1089 singleton pregnancies ended in a livebirth, of which 430 (48%) of 887 pregnancies ended preterm. Each 5 years, we observed more livebirths (RR 1·04, 95% CI 1·01–1·06) and fewer iatrogenic preterm deliveries (0·91, 0·84–0·98). Our data suggest a relationship between platinum-based chemotherapy and small for gestational age (odds ratio [OR] 3·12, 95% CI 1·45–6·70), and between taxane chemotherapy and NICU admission (OR 2·37, 95% CI 1·31–4·28). NICU admission seemed to depend on cancer type, with gastrointestinal cancers having highest risk (OR 7·13, 95% CI 2·86–17·7) and thyroid cancers having lowest risk (0·14, 0·02–0·90) when compared with breast cancer. Unexpectedly, the data suggested that abdominal or cervical surgery was associated with a reduced likelihood of NICU admission (OR 0·30, 95% CI 0·17–0·55). Other associations between treatment or cancer type and outcomes were less clear. Interpretation: Over the years, the proportion of patients with cancer during pregnancy who received antenatal treatment increased, especially treatment with chemotherapy. Our data indicate that babies exposed to antenatal chemotherapy might be more likely to develop complications, specifically small for gestational age and NICU admission, than babies not exposed. We therefore recommend involving hospitals with obstetric high-care units in the management of these patients. Funding: Research Foundation—Flanders, European Research Council, Charles University, Ministry of Health of the Czech Republic.