Following the House of Representatives count , polling officials open and empty the Senate ballot boxes. Ballot papers are unfolded and all the number '1' votes (first preferences) are put into separate piles and counted.

The ballot papers are sorted into first preferences for each party/group (regardless of whether the number 1 is in the group’s above-the-line box or within a candidate’s box below-the-line). Separate piles are created for first preferences allocated to each ungrouped candidate.

After election night, Senate ballot papers initially counted at polling places are securely packaged and delivered to central AEC counting centres where they are received and first preferences are counted again (fresh scrutiny).

These Senate ballot papers are then securely re-packaged and sent to the Central Senate Scrutiny (CSS) site in each state and territory.

The CSS site in each state and territory is where all Senate ballot papers are scanned to capture the millions of further preferences, ready to distribution.

The semi-automated process captures preferences using a combination of optical character recognition software and manual verification by a human operator. Once captured and verified, preferences are stored in electronic tamper‑proof files and progressively published on the AEC tally room.

Once all ballot paper data is received at the AEC it is transferred to the AEC’s Senate counting software where the distribution of preferences is run and final results are determined.

1. Computer recognition (OCR/ICR/OMR) reads the preferences and other marks from the image of a scanned ballot paper.
   
   1a) A data entry operator populates any preferences that were unable to be read with high confidence by the computer recognition in step 1. The data entry operator also manually checks the marks read by computer recognition (for example a voter potentially identifying themselves on a ballot paper) for AEC determination.
   
   *Step 1 and 1a make up the first pass of the data capture process.
2. A separate data entry operator manually enters a complete capture of all preferences marked on the ballot paper and reviews all other marks, this is the second pass of the data capture. The operator at this station cannot see what has been previously captured in step 1 and 1a.
3. If the data from the first pass data capture does not match the data from the second pass data capture the ballot paper is escalated for adjudication by a supervisor. Step 1a and step 2 can escalate to this supervisor for any other reason (e.g. unable to interpret a mark, scrutineer challenge etc.)
4. If a ballot paper is potentially informal or has a sequence break down, the ballot paper is escalated for further adjudication to another human operated queue. If a preference on a ballot paper has been misread, it is likely this will cause a sequence break down or informality, ensuring the ballot paper will be routed to this queue for further analysis.
5. AEC will make any final adjudication on ballot papers and are the only queues authorised to determine a ballot paper is informal for complex escalations.

After each electoral event the AEC completes an assessment of the event, including identifying areas for improvement. From this the systems used are upgraded and fully tested prior to the next election.

EasyCount – Senate (ECS), the Ballot Paper Reconciliation System (BPRS) and the Senate Scanning Solution (SSS) (currently supplied by Fujifilm) all go through testing. This includes unit testing (completed by the developers), function testing (to ensure all components of the system work) and User Acceptance Testing (UAT) to ensure the systems meets the defined requirements. Other, non-functional, testing includes capacity and penetration testing.

Once all systems are built and tested individually, they are tested simultaneously to show they integrate correctly.

Once all systems are shown to accurately work together, the AEC performs further testing of the Senate Scanning Solution in each state and validates the data and flows. After this process is completed, all states process at the same time to ensure the eight (one for each state and territory) production environments operate effectively at the same time.

After the close of nominations, and before polling day, Production Verification Testing ensures the scanners and templates for each ballot paper are correctly set up. Sample ballot papers are used with the resulting data sent from Fujifilm to the AEC, to again test the entire production environment before scanning of senate ballot papers commences, in the days following polling day.

The AEC engage independent providers to demonstrate the software and the Senate scanning processes are accurate. An external provider has demonstrated the software used by the AEC to determine which candidates are elected to the Senate in accordance with the legislation. For more detail, please see the final report: s273AB count software assurance report FE2024/2025.

During the Senate scrutiny process, an external provider will conduct an assurance of the data captured from the Senate ballot papers, in accordance with s 273AC of the act. For detail on the methodology to be used for the ballot paper sampling and the processes to be used for reconciling preferences at the 2025 federal election please see the: s273AC Senate assurance methodology (FE2025).

The final Senate results cannot be calculated until the state or territory-wide total of all votes is known and is used to determine the quota – the proportion of votes required by a candidate to be elected.

It is only possible, therefore, to get an indication of Senate results on election night. While the AEC won’t be able to make a formal declaration for some time after election night it is not uncommon for electoral analysts and others to predict the successful candidates for four or more candidates in a Senate contest from the AEC’s initial publication of first preference results.

The full distribution of preferences will occur up to five weeks after the election.