Background Diminished social networks are common in psychosis but few studies have measured these comprehensively and prospectively to determine how networks and support evolve during the early phase. There is little information regarding perceived support in the early phase of illness. The aim of this study was to describe social support, networks and perceived satisfaction, explore the clinical correlates of these outcomes and examine whether phases of untreated psychosis are linked with social network variables to determine potential opportunities for intervention. Methods During the study period, we assessed 222 people with first-episode psychosis at entry into treatment using valid and reliable measures of diagnosis, positive and negative symptoms, periods of untreated psychosis and prodrome and premorbid adjustment. For follow-up we contacted participants to conduct a second assessment (n = 158). There were 97 people who participated which represented 61% of those eligible. Social network and support information obtained at both time points included the number of friends, self-reported satisfaction with support and social network size and clinician’s evaluation of the degree of support received through networks. Mixed effects modelling determined the contribution of potential explanatory variables to social support measured. Results A number of clinical variables were linked with social networks, support and perceived support and satisfaction. The size of networks did not change over time but those with no friends and duration of untreated psychosis was significantly longer for those with no friends at entry into treatment (n = 129, Median = 24.5mths, IQR = 7.25 – 69.25; Mann-Whitney U = 11.78, p = 0.008). Social support at baseline and at one year was predicted by homelessness (t = -2.98, p = 0.001, CI -4.74 to -1.21), duration of untreated psychosis (t = -0.86, p = 0.031, CI - 1.65 to -0.08) and premorbid adjustment (t = -2.26, p = 0.017, CI -4.11 to -0.42). Social support improved over time but the duration of untreated psychosis was not linked with the rate of improvement in this outcome. Conclusions Improved social support could indicate greater reliance on social support or becoming more adept at mobilising resources to meet social needs. Particularly vulnerable groups with very long duration of untreated psychosis confirm the need for earlier intervention or targeted social network interventions to preserve social connectedness.