The self-medication hypothesis attempts to explain the extraordinary high levels of cigarette smoking in schizophrenia; patients may smoke in an attempt to reduce their cognitive deficits, symptoms, or the side effects of antipsychotics. In a previous report, we detected beneficial performance in attention and working memory in patients with first-episode psychosis who smoked compared to non-smoking patients soon after stabilization. In the present study, we examine differences in the course of those deficits 12 months after the initiation of antipsychotic treatment. We also explore the association between smoking and symptoms and side effects of medication. Neuropsychological assessments were performed at baseline, month 6 and month 12 using a computerized battery that included measures of sustained attention (Continuous Performance Test CPT-O), selective attention (Stroop interference task) and working memory (CPT-XO). Patients met the criterion of fitting in the same smoking category throughout the study: non-smoker (n = 15; 0 cigarettes/day) and smoker (n = 26; >15 cigarettes/day). The non-smoking patients showed significant cognitive improvements, whereas smoking patients lost their superior baseline performance, which was probably obtained through nicotinic stimulation, at the 6- and 12-month assessments due to a static course of deficits. Smokers did not obtain any cognitive benefit after instauration of treatment and worsen their symptoms over the first year. These results suggest that smoking may constitute a marker of a more severe illness. Smoking was not associated with fewer extrapyramidal side effects. Smoking might improve attention and working memory to a similarly modest extent as atypical antipsychotics and could reflect an effort to ameliorate these cognitive dysfunctions previous to treatment instauration.