Fresno’s New Homeless Count Shows Fewer People — But Leaves Many Uncounted

Fresno and Madera counties changed how they count people experiencing homelessness this year, and the new numbers are dramatically lower than before. But there’s a catch: anyone who didn’t agree to answer a survey wasn’t counted at all.

Mayor Jerry Dyer and County Supervisor Nathan Magsig presented the preliminary results Tuesday. The new survey-based approach found 1,619 people unsheltered and 1,635 sheltered — a total of 3,254 people.

That’s a sharp drop from 2025, when the count found 4,905 people experiencing homelessness, up 9.2% from 2023. But this year’s count only included people who voluntarily participated in the survey. Those who declined weren’t added to the tally.

Advertisement

“We are now doing a deeper dive on the individuals who are voluntarily sharing with us how they became homeless, what their needs are, what some of their challenges are,” Magsig said.

In previous years, volunteers with the Fresno-Madera Continuum of Care visually counted everyone they saw experiencing homelessness. This year, they switched to the survey method.

Officials say the new approach will help them better understand what services people need. “This is a more effective way to deal with homelessness because as the population of the homeless changes and their needs change, we need to make sure that our organizations, our services that we provide, change to meet the needs of the people in our community,” Magsig said.

The change comes as California overall saw homelessness decline by nearly 2.8%. Meanwhile, San Francisco and Sacramento did not report figures. And Los Angeles has faced scrutiny for potentially missing up to 7,900 people in their homeless count.

The Breakdown

• Previous counts (2023, 2025): Volunteers visually counted everyone they saw experiencing homelessness.
• 2026 count: Only people who agreed to answer a survey were included in the count.
• 2026 preliminary results: 1,619 unsheltered, 1,635 sheltered — 3,254 total.
• 2025 results: 4,905 people experiencing homelessness (no shelter-status breakdown), up 9.2% from 2023.

What This Means for You

The count you’re seeing isn’t a count of how many people are homeless in Fresno — it’s a count of how many people who are homeless agreed to be surveyed. You’re being asked to trust that the city and county are making progress on homelessness based on a number that, by design, leaves out anyone who didn’t participate. That’s a real shift in how your elected officials are measuring the problem that shapes where public dollars go and whether voters see their investments working. When the methodology changes and the number drops, you have to ask whether the problem got smaller or the measuring stick got shorter.

While having artificially lower figures may make the city look better on paper, having a properly estimated census is important because it provides local organizations with figures that help them to qualify for grants. When the population is underestimated, less funding is on the table, which results in fewer critical services for under-represented populations.

What to Watch

The shift to a survey-based count raises a question officials didn’t answer: how many people did volunteers encounter who declined the survey or couldn’t be reached? That number — the gap between who was counted and who was seen — would tell you whether the drop from 4,905 to 3,254 reflects real progress or just a narrower lens.

There’s also the stewardship question. When the way you measure a problem changes at the same time the number improves, a faithful approach would show both figures side-by-side — what the old method would have found and what the new one did — so the public can see whether the trend is real or whether the methodology itself drove the result. That comparison isn’t in the material. Without it, you’re being asked to take the improvement on faith rather than see it demonstrated.

Is the new survey method genuinely capturing better information to guide services, as officials say, or is it also convenient that it produces a lower count? Those two things can both be true, but only one of them was emphasized at Tuesday’s presentation.