The paper introduces a new method called the Mode-Based Estimate (MBE) for robust inference in summary data Mendelian randomization (MR). MBE aims to estimate a single causal effect from multiple genetic instruments, relaxing the instrumental variable assumptions by assuming that the most frequent value of the bias term (b_j) across all instruments is zero. The method is evaluated through simulations and applied to real data to investigate the causal effect of plasma lipid fractions and urate levels on coronary heart disease (CHD) risk. The results show that MBE performs well under the null hypothesis, with less bias and lower type-I error rates compared to other methods. However, its power to detect a causal effect is smaller than that of the inverse variance weighting (IVW) and weighted median methods. The authors conclude that MBE should be used in combination with other approaches in sensitivity analyses to assess the robustness of causal inference.The paper introduces a new method called the Mode-Based Estimate (MBE) for robust inference in summary data Mendelian randomization (MR). MBE aims to estimate a single causal effect from multiple genetic instruments, relaxing the instrumental variable assumptions by assuming that the most frequent value of the bias term (b_j) across all instruments is zero. The method is evaluated through simulations and applied to real data to investigate the causal effect of plasma lipid fractions and urate levels on coronary heart disease (CHD) risk. The results show that MBE performs well under the null hypothesis, with less bias and lower type-I error rates compared to other methods. However, its power to detect a causal effect is smaller than that of the inverse variance weighting (IVW) and weighted median methods. The authors conclude that MBE should be used in combination with other approaches in sensitivity analyses to assess the robustness of causal inference.