I-485 Final-Stage Median Dips 9 Days, But the 8-Week Climb Continues

The week-over-week drop of 9 days is among the larger single-week moves in the past 12 weeks: only the W35 jump (from 62 to 76 days, a 14-day increase) was larger in absolute terms. The improvement also puts this week modestly below the year-ago reading of 76 days, suggesting that on a year-over-year basis the form is processing somewhat faster. The 12-week arc, however, provides the fuller picture: the median stood at 49 days in late June (week of June 29), climbed to 78 by late August, and has now retreated to 69. The single-week drop does not erase the pattern of the prior two months.

The 8-week regression slope of +2.7 days per week, with an R² of 0.59, indicates a moderate but meaningful upward trend. In statistical terms, R²=0.59 is not a tight fit: week-to-week variance is visible in the table, and any individual week can land several days off the trend line. Still, the direction appears upward, and a slope of this magnitude has accumulated roughly 20 additional days since late July. This week's reading is consistent with normal scatter around that trend rather than a reversal of it.

An RFE (Request for Evidence) is USCIS's standard follow-up letter asking for additional documents. Receiving one does not by itself signal a likely denial; most cases that receive an RFE are ultimately approved. This week, something counterintuitive appeared in the data: cases that had gone through an RFE cycle posted a median of 27 days from last status to approval, against 70 days for cases that processed without any RFE. The automated detector flagged this as statistically unusual relative to the trailing 8-week baseline, producing the rfe_fast anomaly flag.

The speed ratio of 2.61x (non-RFE median divided by RFE median) is the inverse of the usual pattern, where RFE-response cases tend to take longer because of the additional adjudication (USCIS's decision step) review they require. One mechanical explanation is that a group of RFE responses submitted around a similar date cleared adjudication together this week, compressing the final-stage wait for that set. The data brief does not tell us why this occurred, and a single week is not enough to confirm any mechanism. The rfe_response stage across the broader active backlog tells the typical story: 163,518 cases sitting at that status show a median of 122 days, with the middle half (the 25th percentile, p25, to the 75th percentile, p75) spanning 67 to 190 days.

This week, cases that went through an RFE cycle cleared in 27 days, less than half the time of plain-processing cases at 70 days.

Electronically filed cases (routed through IOE) accounted for about 7,001 of 7,830 approvals this week, or 89% of the sample, with a median of 69 days. IOE also carries the most statistically reliable upward slope among centers this week, at +3.1 days per week with R²=0.71, meaning the trend is more consistent here than at any other center. The National Benefits Center (MSC, Lee's Summit MO) posted a median of 41 days across 539 cases this week, a faster pace than IOE, with a steep negative slope of -10.3 days per week. R²=0.45 for MSC limits confidence in that direction, so the apparent improvement should be read cautiously. Case mix differs between these centers, so the comparison is not a straightforward performance ranking.

The California Service Center (WAC) posted a median of 330 days on just 48 cases this week. That sample is too thin to treat as representative: with n=48, weekly variance is large and the figure could shift substantially in either direction with a small change in which cases happened to be approved. The Nebraska Service Center (LIN) posted a median of 145 days across 142 cases, with 8% of those approvals transfer-touched (meaning the case passed through at least one center transfer before reaching LIN). That transfer share means LIN's approved population this week may reflect a different mix of case histories than IOE's or MSC's. Neither gap should be read as a verdict on center-level performance; caseloads and case types differ, and we do not hold data on the destination of transfers or the sub-categories routed to each center.