Background Decreased dietary protein intake (DPI) may lead to protein-energy malnutrition and may be associated with increased mortality risk. We hypothesized that longitudinal changes in dietary protein intake have independent associations with survival in peritoneal dialysis (PD) patients. Methods 668 stable PD patients were selected in the study from January 2006 to January 2018 and were followed up until December 2019. Their three-day dietary records were collected at the baseline (the sixth month after PD) and thereafter every 3 months for two and a half years. The latent class mixed models (LCMM) were used to identify subgroups of PD patients with similar longitudinal trajectories of DPI. The relation between DPI (baseline and longitudinal data) and survival was examined using Cox model to estimate death hazard ratios. Meanwhile, different formulae were used to assess nitrogen balance. Results The results showed that baseline DPI ≤ 0.60g/kg/day was associated with the worst outcome in PD patients. Patients with DPI 0.80–0.99g/kg/day and DPI ≥ 1.0g/kg/day both presented positive nitrogen balance; patients with DPI 0.61–0.79g/kg/day presented obviously negative nitrogen balance. Longitudinal association between time-dependent DPI and survival was found in PD patients. The consistently low DPI' (0.61–0.79g/kg/d) group was correlated with increased death risk as compared with the 'consistently median DPI' group (0.80–0.99g/kg/d, HR = 1.59, p = 0.008), whereas there was no difference in survival between 'consistently median DPI' group and 'high-level DPI' group (≥1.0 g/kg/d, p > 0.05). Conclusion Our study revealed that DPI ≥ 0.8 g/kg/day was beneficial to the long-term outcome for the PD population.