Introduction: Therapeutic plasma exchange (TPE) is used for several chronic conditions with little evidence on the efficacy and safety of different choice of replacement fluid. Measurement of haemostasis, particularly in vitro thrombin generation, could play a role in determining the immediate efficacy of different fluid replacement. Aim: To determine the impact of different TPE replacement fluid regimens on haemostatic assays. Methods: Prospective observational multi-centre cohort study in adult patients 18 years and older evaluating haemostatic changes between four different TPE regimens: (1) 5% human albumin solution (Alb) only, (2) 50:50 mix of 5% Alb + modified gelatin, (3) 70:30 mix of 5% Alb and normal saline (NS), and (4) solvent-detergent, virus-inactivated fresh frozen plasma (FFP) (either alone or combined with other fluids). Twenty-one haemostasis variables were analysed (procoagulant, anticoagulant and fibrinolytic factors) pre and post TPE sessions, including in vitro thrombin generation. Linear mixed modelling and canonical discriminant analyses were used to examine the effect of TPE fluid type on haemostatic variables.Results: A total of 31 patients with up to 5 TPE sessions each (131 sessions in total) were enrolled. Out of 21 markers analysed using linear mixed modelling, the main effects of fluid type were found to be significant for 19 markers (P < 0.05), excluding plasminogen activator inhibitor-1 antigen and thrombinanti-thrombin. Multivariate Analysis of Variance showed significant differences between the fluid types (Wilks' lambda = 0.07; F 63,245.61 = 5.50; P < 0.0001) and this was supported by a canonical discriminant analysis, which identified the 4 most discriminating markers for fluid types as thrombin generation (lag-time, time-to Peak), fibrinogen and Factor V. In our analyses, the effect of FFP on haemostasis was significantly greater compared with other