The instrument that recorded the rise in adolescent distress is now recording a decline, and the direction is harder to argue with than the usual commentary allows. The C.D.C.’s Youth Risk Behavior Survey, fielded in high schools nationwide every two years since 1991, went into the field in 2025. In the data released Monday, 33 percent of students reported persistent feelings of sadness or hopelessness over the previous year, down from a peak of 42 percent in 2021. Fourteen percent reported seriously considering a suicide attempt, down from 22 percent. Attempts in the last year also fell.
Rates had climbed in every cycle since 2011. The reversal first showed up in the 2023 survey; the 2025 numbers extend it.
The comparison that matters most is the one to 2019, before the pandemic. The psychologist Jean Twenge, who has spent years studying teen mental health and smartphone use, calls the declines “very striking, particularly that they’re lower than 2019,” and reads them as something other than an obvious post-Covid rebound: “This isn’t just post pandemic anymore, this is a more longstanding trend in improvement.”
Why the number is worth taking seriously
Survey data of this kind is a blunt instrument. It is self-reported, collected in schools, and fielded every other year — the students most likely to be in crisis, those who have dropped out or are institutionalized, are not in the sample. The questions themselves have changed across iterations.
But it is the same blunt instrument throughout, which is what makes the change in direction informative. Twenge notes that the earlier deterioration showed up in other data sets as well, including clinical depression diagnoses and emergency room visits for self-harm or attempted suicide — so the rise was not an artifact of one questionnaire, and the fall is not obviously one either.
The fall is also not uniform, and the averages conceal most of what a clinician would want to know. Nearly 43 percent of girls reported persistent sadness over the last year, against 24 percent of boys. Nearly 60 percent of students who identified as lesbian, gay or bisexual reported persistent sadness, against 28 percent of heterosexual students; 32 percent had seriously considered attempting suicide, against 11 percent. The overall line bends downward while the groups with the worst outcomes remain at roughly double the rates of their peers. Improvement in the aggregate is not the same as closing the gap, and nothing in these figures suggests the two are moving together.
The rest of the picture
The distress measures are not the only long-running series that keeps falling. Nineteen percent of teenagers reported being sexually active, down from nearly 38 percent in 1991. Twenty percent reported current alcohol use, down from nearly 51 percent in 1991. Fewer than 10 percent reported illicit drug use, against 27 percent when the question was first asked in 2001.
Those trends predate the mood data by decades and are large enough to reframe the story. Adolescent life has been changing along several axes at once — less drinking, less sex, less drug use, more time alone with a screen — and reading the mental-health numbers in isolation flatters whichever explanation the reader arrived with. Thirty-six percent of students said they used social media every hour or more, roughly the same share as in 2023, the first year the question was reported. A variable that stays flat while the outcome falls is not much of an explanation for the fall.
One fewer question
This is the first release of the survey under Health Secretary Robert F. Kennedy Jr., and the questionnaire is not identical to the last one. Data on gender identity, included for the first time in 2023, was removed in 2025 to comply with President Trump’s executive order directing the federal government to recognize only two sexes.
It is worth being precise about what that does. Removing a variable does not change how anyone feels; it changes who can be counted. The groups with the worst reported outcomes in this survey are exactly the ones a coarser instrument describes least precisely — and the next cycle’s ability to say whether those gaps narrowed depends on questions that now do not exist.
What these numbers cannot supply is a cause. The survey establishes levels and direction, not mechanisms, and the honest reading is that the crisis narrative and the improvement narrative are both true of different slices of the same population: the median student is doing measurably better than in 2019, and the students who were always worst off are still carrying most of it.
Read the full story at The New York Times.