Objectives: Pulmonary arterial hypertension (PAH) is a cardiopulmonary disease without effective cure. The objective of this study is to test the hypothesis that PAH patients have a distinct gut microbiome composition acting as a driving force for the increased PAH metabolic biomarkers, acyl carnitines and urate. Methods: Fecal samples from 18 type 1 PAH patients (mean pulmonary arterial pressure [mPAP] 57.4±16.7 mmHg) and 13 reference subjects in Brazil were collected for comparison of microbiome using shotgun metagenomics. Plasma samples from 33 type I PAH patients (mPAP 51.7± 13.5 mmHg) and 19 reference subjects in the USA were used for untargeted metabolomic analysis. Results: More than 80 bacterial taxa were differentially enriched in the microbiomes of the PAH cohort. Random forest machine-learning algorithm predicted PAH patients with 82.4% accuracy from their gut microbiome composition. The most increased taxa in the PAH microbiome were trimethylamine (TMA) producing bacteria like Collinsella (P=0.0037). The PAH metabolome also showed increases of gut microbiome-derived short chain acyl carnitines like isubutyrylcarnitine (P=0.010) and valerylcarnitine (P=0.030), which are the precursors of TMA. In addition, increased urate (P=0.0004) and allantoin (P=0.027) in PAH plasma correlated with significantly increased enzymes producing urate in the PAH microbiome (purine metabolism (P=0.017) and xanthine oxidase (P<0.05)). Conclusion: These findings suggest that the random forest modeling of the gut microbiome composition can be used for prediction of PAH and that the unique gut microbiome of PAH influences acyl carnitines and urate metabolism to contribute to pathophysiology.