Training effective Text-to-SQL models remains challenging due to the scarcity of high-quality, diverse, and structurally complex datasets. Existing methods either rely on limited human-annotated corpora, or synthesize datasets directly by simply prompting LLMs without explicit control over SQL structures, often resulting in limited structural diversity and complexity. To address this, we introduce EvolSQL, a structure-aware data synthesis framework that evolves SQL queries from seed data into richer and more semantically diverse forms. EvolSQL starts with an exploratory Query-SQL expansion to broaden question diversity and improve schema coverage, and then applies an adaptive directional evolution strategy using six atomic transformation operators derived from the SQL Abstract Syntax Tree to progressively increase query complexity across relational, predicate, aggregation, and nesting dimensions. An execution-grounded SQL refinement module and schema-aware deduplication further ensure the creation of high-quality, structurally diverse mapping pairs. Experimental results show that a 7B model fine-tuned on our data outperforms one trained on the much larger SynSQL dataset using only 1/18 of the data. Our code is available at https://github.com/slaanurgle/EvolSQL.

Text-to-SQL is a pivotal task that bridges natural language understanding and structured data access. Recent advances in large language models, through prompting, fine-tuning, and reinforcement learning, have greatly increased the likelihood of generating correct SQL queries. However, selecting the correct SQL from multiple candidates remains challenging. Existing approaches, such as self-consistency, Best-of-N decoding, and learned scalar ranking models, rely on shallow heuristics or lack explicit reasoning, making them prone to unstable judgments and unable to capture fine-grained semantic differences between similar queries. We propose JudgeSQL, a framework that combines reasoning-based SQL evaluation with a weighted consensus tournament. JudgeSQL develops a reasoning-based SQL judge model that distills reasoning traces with reinforcement learning guided by verifiable rewards, enabling accurate and interpretable judgments. Building on this, a weighted consensus tournament integrates explicit reasoning preferences with the generator’s execution-consistency confidence, improving both reliability and efficiency. Experiments on the BIRD benchmark demonstrate that JudgeSQL consistently outperforms prior methods, with strong generalization and robustness across model scales. The code is available at https://github.com/Jiayuan-Bai/JudgeSQL.