Background: The evidence linking low serum sodium levels with the risk of mortality in peritoneal dialysis (PD) patients is controversial. Considering the different mechanisms contributing to hyponatremia in these patients, it is conceivable that the prognostic significance of this factor may vary, according to the clinical setting. Methods: Following a retrospective, observational design, we analyzed the association between hyponatremia and mortality in 748 patients incident on PD. We applied multivariate strategies of analysis, with the main objective of identifying subgroups of patients in whom hyponatremia could sustain different degrees of association with mortality (main outcome variable). For this purpose, we performed preliminary analyses to: (1) disclose predictors of serum sodium levels before and after (mean of first 3 months) initiation of PD (main study variable) and (2) investigate the overall prognostic significance of hyponatremia, in our patients. Results: Comorbidity, hypoalbuminemia, and lower glomerular filtration rate (GFR) were main predictors of hyponatremia. Use of icodextrin was another inverse correlate of serum sodium, and the only consistent predictor of a decline of natremia, once PD was started. Multivariate analysis confirmed early hyponatremia as an independent marker of survival. However, stratified analyses showed that this association was most apparent in specific subsets, namely, hypoalbuminemic, more anemic patients with higher baseline levels of GFR and C-reactive protein and faster peritoneal solute transport rates. Other factors potentially reinforcing the prognostic significance of hyponatremia included lower lean body mass levels, nonprescription of renin-angiotensin-aldosterone system antagonists, and use of icodextrin-based PD solution. On the contrary, baseline overhydration or categorization by classic predictors of mortality (age, comorbidity, diabetes) did not appear to influence the risk pattern associated with lower serum sodium levels. Conclusions: Our results suggest that hyponatremia performs as a consistent correlate of the risk of mortality mainly in PD patients manifesting direct or indirect signs of inflammation and wasting, while this association is not apparently linked to the presence of overhydration or nominal, preexisting comorbid conditions.