We scanned a random sample of 430 residential properties across Baltimore, Maryland and scored each one for visible signs of neglect and disrepair. Baltimore’s rowhouse blocks concentrate distress unevenly, so the citywide average hides a wide neighborhood spread.
Quick answer: In a random sample of 381 scored Baltimore, Maryland homes, 44.9 percent showed at least some visible neglect (95 percent confidence interval 40.0 to 49.9 percent), and the average distress score was 2.45 on a 0 to 5 scale. This is a citywide sample, so it reflects the city itself rather than the wider metro area.
Scored July 2026 with our current production vision model. Part of our multi-city distress study.
How much of Baltimore, Maryland shows visible distress?
Out of 381 homes we could assess from street-level imagery, 171 showed at least some visible neglect, a rate of 44.9 percent. Because this is a random sample rather than a full count, the true citywide rate sits within roughly 40.0 to 49.9 percent at 95 percent confidence. Street-level imagery covered 88.6 percent of the sampled properties.
| Measure | Value |
|---|---|
| Homes sampled | 430 |
| Homes scored | 381 |
| Street-level imagery coverage | 88.6% |
| Some visible neglect (2.5 or above) | 44.9% (40.0 to 49.9%) |
| Clear distress (3.5 or above) | 3.1% |
| Average distress score | 2.45 |
Which parts of Baltimore show the most distress?
Grouping the sample by ZIP code shows where distress concentrates. ZIP is derived from each property’s mapped location, not the owner’s mailing address, so absentee-owned homes count where the house actually sits. Sample sizes per ZIP are small, so read these as directional rather than precise, but the pattern is clear: Southwest Baltimore (21223) is the most distressed area in the sample at 82.6 percent.
| ZIP | Area | Homes scored | Some neglect | Average score |
|---|---|---|---|---|
| 21223 | Southwest Baltimore | 23 | 82.6% | 2.81 |
| 21229 | Irvington | 17 | 64.7% | 2.59 |
| 21213 | Clifton/Belair-Edison | 28 | 60.7% | 2.65 |
| 21217 | Druid Heights | 15 | 60.0% | 2.68 |
| 21206 | Frankford | 25 | 52.0% | 2.42 |
| 21216 | Walbrook | 24 | 50.0% | 2.54 |
| 21218 | Waverly | 32 | 46.9% | 2.54 |
| 21224 | Highlandtown/Canton | 38 | 39.5% | 2.36 |
| 21215 | Park Heights | 32 | 37.5% | 2.38 |
| 21230 | South Baltimore | 40 | 32.5% | 2.31 |
What the imagery actually found
The scoring model records specific named conditions rather than one opaque number, so the distress can be broken down by what is actually visible.
| Condition | Share of scored homes |
|---|---|
| Overgrown or neglected yard | 55.6% |
| Peeling paint or damaged siding | 59.8% |
| Accumulated debris or trash | 30.2% |
| Roof damage | 8.4% |
| Broken or boarded windows | 26.2% |
| Visible structural damage | 5.5% |
| Reads as abandoned | 5.2% |
The common conditions describe wear rather than crisis. The rarer structural conditions are the ones that separate a genuinely distressed house from a merely tired one, and those are a much shorter and more actionable list.
How we did this
- Sample. A random sample of 430 residential parcels drawn from across Baltimore, Maryland, excluding vacant land and government-owned parcels. This is a sample, not a full census of the city.
- City, not metro. These figures describe Baltimore itself. The surrounding suburbs would read differently, and the wider metropolitan county rate would be lower because it folds in more affluent outlying areas.
- Method. For each property we retrieved street-level imagery aimed at the structure and scored it with our current production vision model against a fixed distress taxonomy derived from our own hand-labeled training set. The model returns named indicators with a severity rating plus a distress score from 0 to 5.
- Thresholds. A score of 3.5 or above is clear distress; 2.5 to 3.5 is visible but milder neglect; below 2.5 is unremarkable.
- Confidence. With 381 scored homes, the citywide neglect rate carries a 95 percent confidence interval of about plus or minus 5 percentage points. Per-ZIP figures rest on far smaller samples and are directional only.
- What is not in here. This scan used street-level imagery alone. It includes no aerial imagery scoring and no tax-delinquency data, both of which our platform can add on request.
- Known error mode. Purely visual scoring can over-call distress where heavy shade or dark facades read as damage. We corroborate a visual-only finding against a second signal before treating any single property as a lead.
Common questions
How many distressed properties are in Baltimore, Maryland?
In our July 2026 random sample, 44.9 percent of Baltimore homes showed at least some visible neglect and 3.1 percent showed clear distress. Applied across the city’s housing stock, that points to tens of thousands of properties with visible neglect, though only a full scan would give an exact count.
Which Baltimore neighborhoods have the most distressed homes?
In this sample, Southwest Baltimore (21223) showed the highest rate at 82.6 percent. Several adjacent ZIPs also ran well above the citywide average. Because each ZIP rests on a small sample, treat the ranking as directional.
How current is this data?
The imagery was scored in July 2026. Street-level imagery is itself often one to three years old, so the condition reflects the most recent available photograph of each property rather than today.
How accurate is it?
Accurate enough to rank and prioritize, not to decide on a single address without looking. The model reliably separates a house in poor condition from one in good repair, and it can over-call in unusual lighting, which is why the output is a ranked list to work through rather than a verdict on any one property.
Can I get the underlying property list?
Yes. A full Baltimore scan produces a per-property list with address, owner details, distress score, tier, and the specific conditions found on each home. You can also scan any neighborhood yourself in the app.
Get the Baltimore list
Run your own scan of any Baltimore neighborhood, or request a full-city file. Start scanning, or see plans and pricing. For how these numbers were produced across every city we studied, read the full research writeup.