Calendars, Holidays, and Weather Days

Calendars are the least interesting part of a schedule to set up and one of the most consequential. Every date in the schedule is a duration laid onto a calendar. If the calendar is wrong, every date is wrong, and the error compounds quietly down the critical path. This post covers how many calendars you need, how to handle holidays, the two common ways to account for weather, and the date math that trips people up.

What a calendar does

A calendar tells the scheduling engine which days (and sometimes which hours) count as working time. A 10-day activity on a five-day calendar takes two calendar weeks. The same activity on a seven-day calendar takes ten calendar days. The CPM math does not care about dates at all; it works in units of working time. The calendar is what converts those units back into dates people can read.

Each activity is assigned one calendar. Relationships and lags are also measured against a calendar, and tools differ on which one. Primavera P6, for example, lets you choose whether lags are computed on the predecessor's calendar, the successor's calendar, a 24-hour calendar, or the project default. If you move schedules between tools, check this setting, because a lag of "3 days" can produce different dates depending on it. Lags and Leads covers the rest of the lag behavior.

How many calendars you need

Fewer than you think. Most building schedules work fine with three or four:

Every additional calendar is something someone has to understand and maintain. A schedule with eleven calendars, three of which are nearly identical and one of which nobody remembers creating, is common and is a reliable source of mystery dates. If you import a schedule and find a pile of calendars, consolidate before you baseline.

Holidays

Put holidays on every calendar that should observe them, and decide up front which ones the job will actually observe. That usually means checking the contract, union agreements if any, and asking the superintendent. Typical U.S. construction holidays include New Year's Day, Memorial Day, Independence Day, Labor Day, Thanksgiving and often the day after, and Christmas, with some jobs also shutting down for the week between Christmas and New Year's.

The seven-day continuous calendar usually has no holidays. Concrete cures on Thanksgiving whether or not anyone is on site.

Load holidays for the full duration of the project, not just the current year. A schedule that runs into 2028 with only 2026 holidays will show work on Christmas 2027.

Worked example: the same duration on different calendars

Here is why calendar choice matters. Take two activities that start in mid-November 2026, with Thanksgiving (Thursday, November 26) and the following Friday as holidays on the five-day calendar.

Activity A: 10-day interior framing activity, five-day calendar, starts Monday, November 16.

Week of Nov 16:  Mon 16, Tue 17, Wed 18, Thu 19, Fri 20   -> 5 days
Week of Nov 23:  Mon 23, Tue 24, Wed 25                   -> 8 days
                 Thu 26, Fri 27 are holidays (not counted)
Week of Nov 30:  Mon 30, Tue Dec 1                        -> 10 days
Finish: Tuesday, December 1

Without the holidays loaded, the same activity would finish Friday, November 27, two working days earlier. Small on one activity. Across a critical path that crosses several holiday periods, it adds up.

Activity B: 7-day concrete cure after a slab pour that finishes Wednesday, November 18.

On a 7-day calendar:  Thu 19 through Wed 25  -> finish Wednesday, November 25
On the 5-day calendar: Thu 19, Fri 20, Mon 23, Tue 24, Wed 25,
                      (Thu 26, Fri 27 holidays), Mon 30, Tue Dec 1
                                              -> finish Tuesday, December 1

Same "7 days," six calendar days of difference. If the spec says seven days of cure and means calendar days, putting the cure on the five-day calendar delays every successor for no physical reason. If you want the cure to show as a lag rather than an activity, the same issue applies to which calendar the lag uses.

Weather: two approaches

Weather is the one thing everybody knows will happen and nobody can schedule exactly. There are two common ways to put it in the schedule.

Approach 1: Anticipated weather days on the calendar

You remove an expected number of non-workdays from the weather-sensitive calendar each month, based on historical weather for the site. Exterior activities then automatically take longer in months that are usually bad.

How to get the numbers: look at historical climate data for the nearest station (NOAA publishes historical daily records for U.S. locations), decide what counts as a lost day for your work (rain above some threshold, temperatures below what your concrete or roofing spec allows, high wind for crane work), and count how many such days typically occur per month. Some contracts specify the anticipated weather days per month directly, in which case use theirs.

Strengths: dates are realistic for exterior work, and the baseline already shows the effect of normal weather. Weaknesses: the calendar has to be built carefully, and the specific days you remove are arbitrary since you do not know which Tuesday it will rain.

Approach 2: Weather contingency activity

You keep a normal five-day calendar and add a visible contingency activity, something like "Weather Contingency — 12 days," ahead of a key milestone such as dry-in.

Strengths: simple, visible, easy to explain. Weaknesses: exterior activities show optimistic dates, the contingency sits in one place rather than where the weather actually hits, and there can be disputes about who controls it.

Many contracts dictate which approach to use, usually in the schedule specification. Read Section 01 32 16 and the general conditions before you choose. Reading a Schedule Spec: Section 01 32 16 covers what to look for.

Why the approach matters for delays

The anticipated weather approach sets a clear standard for weather delay claims. If the baseline calendar assumes four lost days in February and you lose nine, the five extra days are the abnormal weather. If you lose three, you have no claim, and you may be ahead.

Without an anticipated baseline, every rainout becomes an argument about whether it was "normal." That is a much harder argument to win or lose cleanly. Whatever method you use, keep daily logs that record weather conditions and whether work was actually stopped on critical activities. Documenting Weather Delays So They Hold Up covers the recordkeeping side. Contracts vary on how weather delays are handled, so read yours and talk to counsel before relying on any particular approach.

Updating calendars during the job

When the job actually loses a day to weather, how you record it depends on the approach:

• With anticipated weather on the calendar: progress the affected activities honestly. If the lost day was one you had already assumed, the schedule already accounts for it. If it was an extra day, it will show up as slippage and you have the data for a request.
• With a contingency activity: reduce the contingency by the days lost, and document it in the update narrative.

Do not quietly edit the calendar mid-job to absorb weather after the fact. It makes the baseline-to-update comparison meaningless, and it is the kind of thing that looks bad in a forensic review.

Common calendar mistakes

• Holidays loaded for only one year.
• Cure and lead times quoted in calendar days placed on a five-day calendar.
• Interior work on the weather calendar, so drywall "loses" rain days.
• Duplicate calendars with tiny differences nobody remembers.
• The project default calendar set to something other than the standard workweek, so new activities pick up the wrong one.
• Saturday work assumed in the calendar that the field is not actually planning to work.

If you inherit a schedule as a P6 .xer or MS Project XML file, review its calendars right after import rather than assuming whoever built them set them up correctly. Moving Schedules Between P6, MS Project, and Everything Else covers what else to check after a conversion.

Key takeaways

• Keep calendars few: standard, seven-day continuous, and weather-sensitive cover most building jobs.
• Load holidays for the whole project duration, and keep them off the continuous calendar.
• Put calendar-day durations (cure, lead times, some review periods) on a seven-day calendar.
• Pick one weather method, match the contract, and stick to it. Anticipated weather days on the calendar make delay claims much cleaner.
• Never edit calendars after the fact to hide weather. Progress honestly and document.