Does training with fewer sets work if you train to muscle failure?
Research shows that fewer sets can work if you train close to muscle failure: for strength the number of sets matters little, and for muscle growth even one set per exercise can be effective.
Fewer sets per exercise can work, as long as you train close to muscle failure. A study in trained men and women found that one set per exercise, twice a week, was already enough to produce measurable gains in strength and muscle growth over eight weeks. The key is not how many sets you do, but how close to your limit you train.
For strength, it makes very little difference whether you train exactly to muscle failure or stop just short of it. Several analyses combining many studies show that the difference between the two approaches is less than 1.3%, which means nothing in practice. This holds across a wide range of repetitions and exercises.
For muscle growth, there is a weak signal that training to or very close to muscle failure produces slightly more than stopping well before it. The effect is small and not visible in all analyses. The relationship is probably not straightforward: beyond a certain point, doing extra repetitions to failure adds very little. Exactly where that point lies is still unclear.
Several reviews conclude that training to muscle failure should be used sparingly. The risk of injury and overtraining increases, though hard numbers are lacking: studies reported very little data on side effects. As a guideline, do not end every set in muscle failure, but do make sure you get close regularly. One well-executed set to just short of failure per exercise is an efficient approach when you are short on time.
This answer draws on several analyses that combined the findings of dozens of studies, supplemented by a study in trained people. The conclusions on strength are fairly consistent and well supported. For muscle growth the evidence is weaker: the effect sizes are small and not always large enough to be considered a real difference. Concrete figures on injury risk are absent; the warnings are based on expert judgement from the reviews, not on measured outcomes.