BACKGROUND: Immune dysfunction has been implicated in negative symptoms of schizophrenia and also in depression. These disorders are frequently co-morbid, with some symptoms such as anhedonia and apathy common to both. The anti-inflammatory agent minocycline may be ineffective in schizophrenia, but more positive effects have been seen in depression. Our aim was to investigate the role of immune dysfunction in depression and sub-domains of negative symptoms in schizophrenia by investigating their intercorrelation and the influence of treatment with minocycline. METHODS: We analysed longitudinal data from 207 patients within 5 years of onset of schizophrenia, from the randomised double-blind, placebo-controlled trial of minocycline (BeneMin). Symptom ratings and circulating IL-6, C-reactive protein (CRP) and TNF-α concentrations were collected at baseline and repeated over twelve months. The sample was not stratified by CRP prior to randomisation. Positive and Negative Syndrome Scale composite ratings of avolition-apathy and diminished expression, Calgary Depression Scale total scores, and immune markers were examined cross-sectionally using Spearman's rank, and longitudinally by linear mixed effect models that included body mass index and minocycline. Additionally, post hoc analysis of the sample stratified by elevated CRP (>1 mg/l and <10 mg/l at baseline) was carried out to assess whether minocycline had any effect on specific symptoms in an immune active sub-group of patients. RESULTS: Depression and avolition-apathy were significantly positively related, and depression correlated weakly with IL-6 at baseline. Diminished expression was associated with increased TNF-α both cross-sectionally and longitudinally. CRP was unrelated to any symptom domain. Minocycline did not affect any individual symptom or sub-domain in the full sample or in the immune active sub-group. DISCUSSION: IL-6 may have some specificity to depression in early schizophrenia. TNF-α may be an indicator of immune dysfunction relevant to negative symptoms, and our longitudinal findings add to this evidence. However, minocycline continues to show very little promise as a treatment for any symptom dimension of early schizophrenia.