Background Erythroderma is a serious condition in itself, quite apart from hazards associated with the underlying disease, and is sometimes fatal. Prognostic studies are rare in the literature and to date there are no published studies to identify the factors that can determine good prognosis. Objectives The aim of the study is to determine the factors that can prognosticate the good outcome of erythroderma. Methodology Cross sectional study from patients diagnosed to have erythroderma between 2003 and 2007 were analyzed with regard to age, sex, race, underlying medical illness, aetiology, duration of rash before diagnosis of erythroderma, response to topical therapy and prognosis. Results Four variables (aetiology, gender, duration of rash before being diagnosed as erythroderma and response to topical therapy) were associated with good prognosis. These variables were statistically significant from univariate analysis. When these variables were included into the binary logistic model, the study did not have enough evidence to proof that ‘aetiology’ and ‘gender’ can determine good prognosis. Response to topical therapy and shorter duration of rash (equal and less than 120 days) were significant with odds ratio (CI) of 4.11 (1.556, 10.885) and 4.608 (1.903, 11.155) respectively. Conclusion Shorter duration of rash and response to topical therapy are important factors to determine a good prognosis.