OBJECTIVES: The evidence on the influence of timing of discharge on readmission risk is inconsistent, showing that both increased length of stay (LOS) and early discharge are associated with an increased risk of readmission. We aimed to test whether LOS and day of discharge (i.e., weekend vs. work days) are associated with the risk of 30-day emergency readmission. METHODS: Data on first admission to internal medicine departments during January – March 2010 was retrieved from Clalit Health Services' data warehouse. Inclusion criteria: LOS of >= 2 nights, age 18+, readmitted to an internal medicine demartment or ICU. Predictors: LOS, ACG morbidity categories, age, socioeconomic status, and prior healthcare use. Logistic regression was used to model the effect of discharge day and LOS on 30-day emergency readmission, controlling for known risk factors. RESULTS: After adjustment for morbidity, clinical and demographic factors, there is an increased risk for 30-day readmissions associated with increased length of stay (OR=1.56 when LOS is 8+ days vs. 2-3 days, p<0.001). Focusing on 2-7 day hospitalizations, the same association is found (OR=1.47 for LOS > 5 days, p<0.001). Being discharged over the weekend increases the odds of readmission by 11% (p=0.04), controlling for all known risk factors. Modeling for an outcome of readmission or death, or for a 7-day readmission outcome, resulted in similar findings. CONCLUSIONS: Our study showed that contrary to some of the evidence, longer hospital stays were associated with increased risk of unplanned readmission. Being discharged over the weekend incurs mild additional risk for readmission, especially at shorter LOS. These findings suggest that despite short LOS in Israel, increasing LOS as a readmissions reduction strategy may be ineffective, unless the specific need and care content of these additional days is assured.