I-765 Weekly Median Rises 29 Days, But the 8-Week Trend Still Points Down

The I-765 final-stage wait (median days from last reported status to approval) rose to 140 days in the week of September 14, up 29 days from 111 days the prior week. The immediate instinct is to read this as a reversal, but the 8-week regression slope tells a different story: at -8.2 days per week, the underlying trend is the most sustained improvement in the 12-week window. The median peaked at 198 days in early August (week of August 3) and has broadly declined since, a drop of roughly 58 days over eight weeks. At a sample of 11,584 approved cases this week, the median is statistically reliable; the week-over-week movement reflects composition as much as operational change.

One-week moves in the median can reflect the composition of which cases happen to clear in a given week, not necessarily a policy or operational shift. This week's spread between the 25th percentile (p25, meaning a quarter of approved cases finished within this time) at 78 days and the 75th percentile (p75, meaning three-quarters finished within this time) at 257 days illustrates how mixed the approved population is: a 179-day interquartile range (the spread from p25 to p75) means very different case types cleared together. Year over year, the 140-day median sits 15 days above the same week in 2025 (125 days), a gap worth tracking but well within the volatility the 12-week table shows.

The 12-week trend table runs from a 100-day reading in late June to a 198-day peak in early August, then a broad decline toward this week's 140 days. The 8-week regression slope of -8.2 days per week translates to roughly 66 days of improvement across that span, the fastest sustained recovery in this data window. The R² of 0.46 is the important qualifier: the trend is directionally real, but the fit is moderate, meaning individual weeks can deviate substantially from the regression line. The W36-to-W37 drop of 52 days and the W32-to-W33 drop of 60 days both illustrate how much a single week can move in either direction within the same improving trend.

If the recent pace holds, the week of September 21 projects to approximately 84 to 180 days, narrowing toward 60 to 155 days by mid-October. These bands are wide because the underlying R² of 0.46 does not support a single-number projection: the regression describes a direction, not a precise endpoint. The central message is directional rather than precise. The slope points toward shorter waits, but any given week can still land anywhere across a roughly 90-day window.

The trend line points consistently downward, yet any single week can land dozens of days from where the regression expects.

An rfe_fast anomaly flag fired this week, capturing an inversion worth examining. Cases that received and responded to a Request for Evidence (RFE, USCIS's standard follow-up letter asking for missing documents) cleared in a median of 59 days, compared with 140 days for cases that never received one. That is a 2.37x speed ratio in favor of RFE-stamped cases, the opposite of the typical pattern, where RFE response cases routinely take 30 to 90 days longer than non-RFE cases. The most likely explanation is a batch of older RFE-response cases that cleared together in a single week, compressing that group's median. The data brief does not confirm this mechanism, but batched clearance is consistent with a one-week signal of this type.

Stage durations data (30-day raw window, a separate dataset from the weekly aggregate) shows cases at the 'Response To USCIS Request For Evidence Was Received' status with a median of 39 days, a p25 of 16 days, and a p75 of 90 days across 6,616 cases. That wide spread from 16 to 90 days is consistent with batched clearance pulling some cases through quickly while others wait. Attorneys should treat this week's rfe_fast signal as a one-week observation: the structural expectation remains that RFE response review adds weeks or months to total processing time.