Project Network Diagram: How to Draw One and Read Float

A project network diagram is a schedule with the calendar removed: boxes for activities, arrows for dependencies, and the float written on each box. The idea is not new. The Critical Path Method that reads these diagrams was created in the late 1950s by Morgan Walker of DuPont and James Kelley Jr. of Remington Rand, and the layout has barely changed since.

The direct answer: to draw a project network diagram, put each activity in a box, draw an arrow from each predecessor to its successor, then run a forward and a backward pass to write earliest and latest dates on every box. The chain with zero float is the critical path. Everything else has slack you can read straight off the diagram.

In short. A network diagram is activity-on-node: one box per activity, one arrow per dependency. It answers "why is the finish date what it is", which a Gantt chart cannot show as clearly. Draw it, run the two passes, and the critical path and the float on every other activity fall out.

What is an activity-on-node diagram?

Activity-on-node (AON) is the standard layout for a project network diagram: activities are nodes, dependencies are the arrows between them. The precedence diagramming method is the same thing under another name. Older texts use activity-on-arrow, where the arrows are the work, but almost no current tool draws it that way.

Each box carries the same few numbers:

Arrows carry the dependency type (finish-to-start, start-to-start, finish-to-finish or start-to-finish) and any lag. The four types are covered in dependency types explained.

How do you draw a project network diagram?

Draw it in five steps:

1. List the activities with durations. Eight is enough to learn on.
2. Record each activity's predecessors. Ask of each one: what must finish before this can start?
3. Place boxes left to right so every arrow points forward, with activities that have no predecessor in the first column.
4. Draw the arrows, labelling any that are not finish-to-start or that carry a lag.
5. Run the passes and write ES, EF, LS, LF and float in each box. The arithmetic is worked step by step in the critical path method explained.

Here is a small plan. Durations are in working days, every link is finish-to-start, and day numbers are offsets from the project start.

The diagram below shows this plan with the critical path in red.

How do you read the critical path and float off the diagram?

The critical path is the unbroken chain of boxes with zero total float, and in this plan it is A, B, C, E, G, H, which gives a 31-day project. Total float is the latest start minus the earliest start; any activity above zero can slip that many days without moving the finish.

The two blue boxes are the useful part. Procurement (D) and environment setup (F) take 12 days together, against 16 days for design plus build running alongside, so each has 4 days of total float. A 6-day slip on D would push the finish by 2 days; a 3-day slip would change nothing.

Float comes in two kinds, and the diagram shows both. D has 4 days of total float but zero free float, because F starts the moment D finishes, so any slip on D moves F immediately. F has 4 days of free float, because G cannot start until day 23 whatever F does. The difference is explained in total float vs free float.

The 2-day lag between G and H is why Go-live starts on day 30, not day 28. Lag sits on the arrow, not in a box, so a diagram is the clearest place to check that nobody has hidden a wait inside a duration.

When is a Gantt chart the better view?

Use a Gantt chart when the audience wants dates, and a network diagram when you are checking logic. A network diagram answers "why is the finish date what it is"; a Gantt chart answers "what is happening in March". Most tools draw both from the same data. Microsoft Project has a Network Diagram view on its View tab, and in Onplana the Gantt carries the same dependencies and critical path on every plan, including Free. For reading and editing the timeline, see the Microsoft Project Gantt chart explained and what a Gantt chart is. The difference between CPM and its probabilistic cousin is in PERT vs CPM.

What goes wrong in real network diagrams?

Real schedules break the diagram in three predictable ways:

• Dangling activities. A box with no successor has no effect on the finish date, so it can slip forever without showing up.
• Missing predecessors. An activity with no predecessor starts on day 0 on paper, which understates the real critical path.
• Lag used as a plug. A "+5d" on an arrow is often an estimate hiding in the logic. Write the wait as its own activity so it can be questioned.

Each of these is a structural check that a tool can run on a file you already have. The seven checks to make on a schedule first cover them, and the free Schedule Health Check reads a file before you commit to anything. If you keep working the schedule afterwards, the /schedule page covers Onplana Schedule, the desktop app.

Run the free Schedule Health Check Upload a schedule and see dangling tasks, missing links and constraint conflicts, the structural faults that make a network diagram lie. No signup required. → Open the Schedule Health Check

Microsoft Project™ is a trademark of Microsoft Corporation. Onplana is not affiliated with Microsoft.