In recent years, studies on the clinical features and cognitive impairment of patients with different first-episode types of bipolar disorder have received increasing attention. The patients with bipolar disorder may present with different symptoms at first onset. The aim of this study is to assess the cognitive functions of a patient's index episode of bipolar disorder, depression or mania, on risk factors of effecting on cognitive functions. One hundred sixty eight patients with bipolar disorder diagnosed for the first time were enrolled in the study. All patients were divided into two groups according to their index episode of bipolar disorder, either depression or mania. Seventy three patients of the cohort had an index episode mania and 95 patients had initial symptoms of depression. Demographic and clinical disease characteristic data of all enrolled patients were collected. Meanwhile, 75 healthy controls were included. Demographic data of controls were collected. The cognitive functions of all patients and controls were detected by continuous performance test (CPT), digital span test (DST) and Wisconsin card sorting test (WCST). The main cognitive functions data were compared among the mania group, depression group and control group. The relevant risk factors affecting cognitive function were analyzed. (1) Most patients with bipolar disorder had an index episode depression (56.55% vs. 43.45%). Compared with the depression group, the mania group had later age of onset [(24.01 ± 4.254) vs. (22.25 ± 6.472), t = 2. 122, p = 0.035]. The education level of patient groups was lower than control group (p < 0.001). (2) The healthy control group's DST, WCST and CPT scores were better than the patient groups (All p < 0.05). The mania group's DST (forward, reverse, sum), WCST (total responses, completed classifications, correct responses, incorrect responses, percentage of correct responses, completed the number of responses required for classification, the percentage of conceptualization level, the number of persistent responses, non-persistent errors), CPT (2 digit score, 3 digit score, 4 digit score) was better than the depression group (p < 0.05). (3) In mania group, correlation analysis showed that all CPT parameter, inverse digit span, and the sum of DST was negatively correlated with the education level (All p < 0.05). The CPT-4 digit score was negatively correlated with onset age (p < 0.05). In the WCST, the number of correct responses, the percentage of correct responses and the percentage of conceptualization level were positively correlated with the BRMS score (All p < 0.05). The number of false responses and persistent responses were negatively correlated with the BRMS score (All p < 0.05). The number of persistent errors and percentage of persistent errors was positively correlated with education years (All p < 0.05). In depression group, there was a positive correlation between inverse digit span and the education level (p < 0.05). In our study, there were cognitive impairments in attention, memory, and executive function of patients with different onset syndromes of bipolar disorder. Compared with the mania group, the degree of cognitive impairments in bipolar patients with the depressive episode was more severe. The risk factors affecting cognitive impairments included the age of onset, education level, number of hospitalizations and severity of illness.