Background: Although the risk of suicidality is high in first-episode psychosis, patterns and individual variability in suicidal thoughts and behaviours over time are under-researched. We aimed to identify early trajectories of suicidality over a 2-year follow-up, assess their baseline predictors, and explore associations between those trajectories and later suicidality. Methods: This longitudinal follow-up study was a part of the Early Treatment and Intervention in Psychosis (TIPS)study. Participants, linked to Norwegian and Danish death registries, were recruited from four catchment areas (665 000 inhabitants) in Norway and Denmark (both inpatient and outpatient). We included participants aged 15–65 years, with an intelligence quotient of more than 70, willing to give informed consent, and with a first episode of active psychotic symptoms. Individuals with comorbid neurological or endocrinal disorders, or those with contraindications to antipsychotics, were excluded. Growth mixture modelling was used to identify trajectories of suicidal thoughts and behaviours over the first 2 years. Multinomial logistic regression was applied to examine the baseline predictors of those trajectories and their associations with suicidality at 10-year follow-up. Findings: A total of 301 participants were recruited from Jan 1, 1997, to Dec 31, 2000. Of the 299 with completed suicidality data at baseline, 271 participated in 1-year follow-up, 250 in 2-year follow-up, 201 in 5-year follow-up, and 186 at 10-year follow-up. At baseline, 176 (58%) were male, 125 (42%) were female. The mean age was 27·80 years (SD 9·64; range 15–63). 280 (93%) participants were of Scandinavian origin. Four trajectories over 2 years were identified: stable non-suicidal (217 [72%]), stable suicidal ideation (45 [15%]), decreasing suicidal thoughts and behaviours (21 [7%]), and worsening suicidal thoughts and behaviours (18 [6%]). A longer duration of untreated psychosis (odds ratio [OR] 1·24, 95% CI 1·02–1·50, p=0·033), poorer premorbid childhood social adjustment (1·33, 1·01–1·73, p=0·039), more severe depression (1·10, 1·02–1·20, p=0·016), and substance use (2·33, 1·21–4·46, p=0·011) at baseline predicted a stable suicidal ideation trajectory. Individuals in the stable suicidal ideation trajectory tended to have suicidal thoughts and behaviours at 10-year follow-up (3·12, 1·33–7·25, p=0·008). Individuals with a worsening suicidal trajectory were at a higher risk of death by suicide between 2 and 10 years (7·58, 1·53–37·62, p=0·013). Interpretation: Distinct suicidal trajectories in first-episode psychosis were associated with specific predictors at baseline and distinct patterns of suicidality over time. Our findings call for early and targeted interventions for at-risk individuals with persistent suicidal ideation or deteriorating patterns of suicidal thoughts and behaviours, or both. Funding: Health West, Norway; the Norwegian National Research Council; the Norwegian Department of Health and Social Affairs; the National Council for Mental Health and Health and Rehabilitation; the Theodore and Vada Stanley Foundation; the Regional Health Research Foundation for Eastern Region, Denmark; Roskilde County, Helsefonden, Lundbeck Pharma; Eli Lilly; Janssen-Cilag Pharmaceuticals, Denmark; a National Alliance for Research on Schizophrenia and Depression Distinguished Investigator Award and The National Institute of Mental Health grant; a National Alliance for Research on Schizophrenia & Depression Young Investigator Award from The Brain & Behavior Research Foundation; Health South East; Health West; and the Regional Centre for Clinical Research in Psychosis.