In computer science, a conference's acceptance rate functions as a shorthand for prestige — but the story behind the numbers is considerably more complex than the single percentage suggests. Understanding what acceptance rates actually measure, how they have evolved, and what they mean for submission strategy can save researchers significant time and frustration.
What Acceptance Rate Means and How It Is Calculated
Acceptance rate is simply the number of accepted papers divided by the number of submissions received, expressed as a percentage. However, this apparently simple calculation conceals important variation. Some venues count all submissions including desk-rejected or withdrawn papers in the denominator; others count only papers that completed full review. Some report acceptance rates that include workshop papers or short papers alongside full research papers; others report figures only for the main track. When comparing acceptance rates across venues or across years, always check the methodology behind the number before drawing conclusions.
Acceptance Rates at Top CS Venues: The Approximate Picture
The most selective top-tier venues in computer science consistently report acceptance rates in the 15–25% range, though with considerable variation by year. NeurIPS (Neural Information Processing Systems) has historically ranged from roughly 15% to 26% across recent years, with the variance partly reflecting changes in review processes and submission volumes. ICML (International Conference on Machine Learning) typically accepts around 18–22% of submissions. CVPR (Computer Vision and Pattern Recognition) generally operates in the 24–28% range, reflecting its large submission volume. In natural language processing, ACL tends toward 20–25% for the main conference. In software engineering, ICSE (International Conference on Software Engineering) is highly selective, with acceptance rates frequently in the 16–22% range for the technical track. These figures are approximate and shift annually — always consult the most recent program statistics published by the organizing committee.
Why Low Acceptance Rate Does Not Always Equal Better Quality
The intuition that lower acceptance rates signal higher quality is widespread but imperfect. Acceptance rate is a function of both the quality of accepted papers and the volume and quality distribution of submissions. A venue that attracts a large number of speculative or poorly matched submissions will appear highly selective even if the review bar is unchanged. Conversely, a venue with a higher acceptance rate may maintain extremely high quality if it attracts only serious, well-developed work from experienced authors. Review quality, area chair calibration, reviewer expertise, and the presence of rebuttal processes are better proxies for quality control than the acceptance rate percentage alone.
How Acceptance Rates Have Changed Over Time
The most dramatic trend in CS conference acceptance rates is the explosion in submission volume at machine learning venues. NeurIPS received fewer than 1,000 submissions as recently as 2014; by the early 2020s, submission counts had grown to more than 10,000 per year. When acceptance rates stayed roughly constant, the absolute number of accepted papers grew enormously — raising questions about whether the review infrastructure could maintain calibration at scale. Many venues have responded by expanding area chair pools, introducing more structured review guidelines, experimenting with reviewer bidding reforms, and in some cases introducing submission caps or two-stage review processes. The trend reflects the extraordinary growth of machine learning as a field, not a degradation of quality standards.
What Acceptance Rate Should You Target When Choosing Where to Submit?
The most important strategic consideration is matching the maturity and contribution level of your work to the appropriate venue. Submitting under-developed work to a highly selective venue in hopes of getting detailed feedback is generally a poor trade — the rejection rate is high and turnaround times at top venues can run four to six months. A better framework considers: the topical fit of your work to the venue's community, the stage of your research (early-stage work may benefit from workshop submission first), the timeline pressure on your career or project, and the review culture of the venue (some venues are known for constructive reviews even on rejected papers). Acceptance rate is a secondary factor once fit and readiness are established.
How Reviewers Affect Acceptance Rates
Reviewer assignment, expertise matching, and the presence of rebuttal processes significantly affect which papers are accepted. Studies of major CS conferences have documented substantial inconsistency in reviewer judgments — papers rejected at one top venue are frequently accepted at a venue of comparable prestige. The NeurIPS 2014 experiment, in which a subset of papers were reviewed twice by independent committees, found that roughly half of accepted papers would have been rejected under the alternate review assignment. This finding has been replicated informally in subsequent years. The implication for submitters: a rejection from a top venue is meaningful feedback, but it is not a definitive judgment of the work's quality or eventual impact.
Using Acceptance Rate Data Wisely in Your Submission Strategy
Acceptance rate data is most useful as one input into a broader submission strategy. Track rates over multiple years at your target venues to identify trends — a venue whose acceptance rate is declining while submissions are growing may be entering a more competitive phase. For junior researchers, it is worth targeting a portfolio of venues across selectivity tiers in parallel, when conference timelines permit. Resources such as LatestConferences.com aggregate upcoming conference deadlines and can help you identify the right submission windows across venues in your field, so that you can plan a realistic submission calendar rather than reacting to individual deadlines in isolation.
Beyond Acceptance Rate: Other Quality Signals
Acceptance rate deserves to be read alongside other quality indicators: the impact factor or h-index of proceedings in your field's citation databases, the breadth and seniority of the program committee, the quality of the best paper awards over recent years, and community reputation as evidenced by where the most-cited recent papers in your area were first published. A venue with a 30% acceptance rate and rigorous double-blind reviewing by a well-calibrated community may serve your career better than a 10% acceptance rate venue with known reviewing inconsistencies. Use all available evidence.