Abstract Background Despite advances in mapping and ablation techniques, atrial fibrillation (AF) recurrence remains a challenge after pulmonary vein isolation. Low voltage areas (LVA) in the left atrium are associated with higher recurrence rates after left atrial ablation. Risk scores for evaluating post-ablation recurrences are not well established, although a better patient selection for repeat procedures would be of great help in clinical practice. Purpose Risk scores for prediction of AF recurrence have not been tested in a cohort of repeat ablations. Methods This single-center study includes consecutive patients from the prospective Bernau ablation registry undergoing ultra-high-density (UHD) mapping and repeat ablation for AF/AT recurrence between 2016 and 2020. The potential of seven guideline mentioned risk scores (APPLE, DR-FLASH, MB-LATER, ATLAS, CAAP-AF, BASE-AF 2, ALARMEc) to predict (1) AF/AT recurrence beyond a three-months blanking period after repeat left atrial ablation (Re-PVI ± further LA ablation) and (2) the percentage of left atrial LVA in UHD mapping was investigated. LVA were defined as sites with a bipolar peak-to-peak voltage of <0.5 mV with an extent of >1cm². Optimal cutoff for sensitivity and specificity for LVA and AF/AT recurrence as endpoint was chosen using C statistics with receiver-operator characteristics (ROC). Further ROCs were performed to illustrate the predictive ability of the scores. Pearson correlation was used to test associations between variables. Results 160 patients (mean age 67.9 ± 9.1 years, 60.6% persistent AF, mean AF duration 4.6 ± 3.8 years) with complete left atrial UHD mappings (mean EGMs 9754 ± 5808) were included. Overall recurrence rate over a mean follow-up time of 16 ± 11 months was 55.6%. The predictive value of the investigated risk scores on AF/AT recurrence in our cohort was low (Table 1), with the highest power for CAAP-AF (p = 0.015, AUC = 0.615) and DR-FLASH score (p = 0.040, AUC = 0.594), Figure 1. With respect to left atrial LVA we found a better predictive power for the CAAP-AF (p < 0.001, AUC 0.702), APPLE (p < 0.001, AUC 0.687), DR-FLASH (p < 0.001, AUC 0.688), ATLAS (p = 0.005, AUC 0.634) and ALARMEc (p = 0.007, AUC 0.608) score to predict low voltage based on a calculated cut-off of 22% of total left atrial surface (Figure 1). Conclusion The predictive value of guideline-referred risk scores in estimating AF/AT recurrence after repeat ablation is low and does not seem to be of relevant help in patient selection for further interventional treatment. Some scores demonstrate a fairly good prediction for the amount of left atrial LVA and therefore might help in choosing the right mapping and ablation regime beforehand.