I-90 Final-Stage Wait Doubles to 104 Days in Preliminary Early Read
The week of Aug 17 produced a preliminary median of 104 days, up 58 days from the prior week and more than double the 12-week trailing average of 41 days. The anomaly detector fired at z=11.62 against the prior 4-week trailing trend, the most extreme single-week departure in the 12-week data window.
A preliminary read for the week of Aug 17 puts the I-90 final-stage wait at 104 days, up 58 days from last week's 46 and more than double the 12-week trailing average of 41 days. The sample stands at 3,856 approved cases, roughly 50% to 70% of the prior weekly pace, so the number carries real uncertainty. Even so, the magnitude of the move is large enough that sample composition alone is unlikely to explain it fully. The anomaly detector fired at z=11.62 against the prior 4-week trailing mean, the sharpest reading in the 12-week window. An 8-week slope of +9.6 days per week (R²=0.76) shows the trend was already bending upward before this week's reading arrived.
- This week's preliminary median of 104 days is +58 days above last week's 46 days, a +124.6% week-over-week move against a 12-week trailing average of 41 days.
- The outlier anomaly detector fired at z=11.62 against the prior 4-week trailing trend, the most extreme single-week departure recorded in the 12-week data window.
- The 8-week regression slope of +9.6 days per week (R²=0.76) shows a strong pre-existing acceleration that predates this week's jump.
- RFE (Request for Evidence)-stamped cases cleared in a median of 19 days this week, faster than the 33-day median for plain-processing cases — an unusual inversion flagged by the rfe_fast anomaly.
- The overall active backlog across all filing-month groups fell by about 8,300 cases (9.8%) over the 28 days ending Aug 26, indicating that adjudications continued even as the final-stage wait climbed.
104 days: a spike that demands context
The I-90 final-stage wait came in at 104 days for the week of Aug 17, up from 46 days the prior week and 63 days above the 12-week trailing average of 41 days. This is an early, preliminary read: the sample stands at 3,856 approved cases, roughly 50% to 70% of the prior weekly volume, which introduces meaningful uncertainty into the single-week figure. The anomaly detector fired at z=11.62 against the prior 4-week trailing mean, the most extreme departure in the 12-week data window. That score reflects how far this week's reading sits above the recent run-rate, not a comparison to all historical data.
Against the same week last year, the 104-day preliminary figure is +11.8% above the 93-day reading for W34 2025 (n=2,688). That year-over-year gap is real but considerably smaller than the week-over-week shock, and last year's same-week figure was itself above what was then a normal baseline. The context matters: this week's reading is elevated on both a short-term and a year-over-year basis, though the near-term move is the more pronounced of the two.
A number this far above the recent run-rate is worth tracking closely, but the preliminary sample means confirmation must wait for next week's full count.
A trend that was already bending upward
This week's reading did not emerge from flat ground. The 8-week regression slope stands at +9.6 days per week, with an R² of 0.76, indicating a strong and consistent upward drift since late June. The median was 15 days in W25 (Jun 15), rose to 36 days in W29 (Jul 13), reached 53 days in W32 (Aug 3), and now sits at a preliminary 104 days in W34. At the current slope, the trend line would project something in the range of 65 to 75 days for this week — this week's 104-day reading sits well above even that accelerating baseline, pointing to a discrete additional factor layered on top of the underlying drift.
The interquartile range, the spread between the 25th percentile (p25) and the 75th percentile (p75), also widened sharply. This week's band runs from p25=60 days to p75=111 days, an IQR of 51 days. In W32 (Aug 3), the same band ran from p25=10 days to p75=54 days — far tighter. A widening spread alongside a rising median is consistent with heterogeneity among the cases clearing this week: different types of cases, or cases from different filing periods, appear to be mixing in the approved pool rather than a uniform shift in the processing timeline.
Where pending cases sit right now
The filing-month cohort (a group of cases) burndown data offers a complementary view. Cases filed in April 2026 (age 21 weeks) are 53% approved as of the Aug 26 snapshot; those filed in June 2026 (age 12 weeks) are 43% approved; cases filed in July 2026 (age 8 weeks) are 28% approved. These cohort percentages measure a different thing from the weekly median: they track what share of each filing-month group has cleared over its lifetime, not the final-stage wait of cases approved in any given week. Separately, the total active pending count across all filing-month groups fell by about 8,290 cases (9.8%) over the 28 days ending Aug 26, a sign that the adjudication (the USCIS case decision step) pipeline kept moving despite the rise in final-stage wait times.
Preview
Unlock the full analysis for your case
See the complete outlook, a personalised lookup against this week's cohort, and your estimated approval window.
- Full premium analysis: outlook, what it means for you, and implications.
- Personalised case lookup against this week's cohort, with percentile and remaining days.
- Estimated approval window mapped to your filing date.
Cancel anytime. Plans from $5.99/month.