Australians lose more money gambling, per person, than any nation on Earth. This is the story of how much we lose, what we lose it on, why no two states gamble alike, and who quietly profits.
No country loses more money gambling per person than Australia, and we keep losing more. Back in 1975–76, Australians lost about A$884 million in total. By 2023–24 that had grown to A$32.2 billion.
Part of that jump is just more people and rising prices over fifty years. That’s why the later charts switch to a per-person view, to take population out of the picture. But even then, the line keeps going up. So the real question is this: where does all that money go, and who gets it?
aus_total <- read_excel(exp_file, skip = 7,
col_names = c("year","spacer","NSW","Vic","Qld","SA","WA","Tas","NT","ACT","Australia")) %>%
select(-spacer) %>%
mutate(across(NSW:Australia, as.numeric)) %>%
filter(!is.na(Australia)) %>%
mutate(year_start = as.integer(str_sub(year, 1, 4)),
aus_billion = Australia / 1000)
plot_ly(aus_total, x = ~year_start, y = ~aus_billion,
type = "scatter", mode = "lines",
line = list(color = "#d8352a", width = 3),
customdata = ~year,
hovertemplate = "%{customdata}<br><b>A$%{y:.1f} billion</b><extra></extra>") %>%
conv_layout("Australia's gambling losses keep climbing",
"Total losses (A$ billion, nominal)", legend = FALSE)Figure 1: Total gambling losses in Australia, 1975–76 to 2023–24 (nominal A$). Source: Queensland Government Statistician’s Office (2025).
Pokies are by far the biggest. In 2023–24 we lost A$16.3 billion on gaming machines alone, more than wagering, casinos, lotteries and keno put together. The small dip around 2019–20 is COVID, when venues had to close. After that the line bounced back to a new high.
The fastest-growing one is wagering, which means betting on racing and sport. It has more than doubled since the mid-2010s, from about A$3.5 billion to A$8.4 billion, mostly because betting moved onto phones and apps. Casino losses have actually fallen since their peak. Lotteries grow slowly, and keno stays tiny. Hover over any line to see a single year.
prod_exp <- c("Gaming machines (pokies)" = "GAMING_MACHINES 5",
"Wagering (racing & sports)" = "WAGERING 5",
"Casino" = "CASINO 5",
"Lotteries" = "LOTTERIES 5",
"Keno" = "KENO 5")
chart2_data <- imap_dfr(prod_exp, ~ read_ags(prod_file, .x) %>%
filter(state == "Australia") %>%
mutate(product = .y)) %>%
mutate(product = factor(product, levels = names(prod_exp)),
value_b = value / 1000)
plot_ly(chart2_data, x = ~year_start, y = ~value_b, color = ~product, colors = pal5,
type = "scatter", mode = "lines", line = list(width = 2.5),
hovertemplate = "%{fullData.name}<br>%{x}: A$%{y:.2f}b<extra></extra>") %>%
conv_layout("Pokies still rule \u2014 but betting is surging",
"Expenditure (A$ billion, nominal)")Figure 2: National gambling losses by type, over time (nominal A$). Source: Queensland Government Statistician’s Office (2025).
Once you look at losses per adult, New South Wales comes out on top: around A$2,000 per person in 2023–24, and most of that goes into pokies. That makes sense, since NSW has more poker machines than almost anywhere outside Nevada.
The odd one out is Western Australia. It’s the only state where pokies barely show up at all, because WA bans them everywhere except the Perth casino. So people in WA bet on racing and buy lottery tickets instead. It’s a neat real-life experiment: take the machines away, and a whole state ends up gambling differently.
prod_pc <- c("Pokies" = "GAMING_MACHINES 7",
"Wagering" = "WAGERING 7",
"Casino" = "CASINO 7",
"Lotteries" = "LOTTERIES 7",
"Keno" = "KENO 7")
chart3_data <- imap_dfr(prod_pc, ~ read_ags(prod_file, .x) %>%
filter(year_start == 2023, state != "Australia") %>%
mutate(product = .y)) %>%
mutate(product = factor(product, levels = names(prod_pc)))
state_order <- chart3_data %>% group_by(state) %>%
summarise(t = sum(value, na.rm = TRUE)) %>% arrange(desc(t)) %>% pull(state)
chart3_data <- chart3_data %>% mutate(state = factor(state, levels = state_order))
plot_ly(chart3_data, x = ~state, y = ~value, color = ~product, colors = pal5,
type = "bar",
hovertemplate = "%{fullData.name}<br>%{x}: A$%{y:.0f}/person<extra></extra>") %>%
layout(barmode = "stack") %>%
conv_layout("Per-person losses by gambling type, 2023\u201324",
"Expenditure per adult (A$)", xtitle = "")Figure 3: Losses per adult by gambling type and state, 2023–24 (A$ per adult). Source: Queensland Government Statistician’s Office (2025).
For about twenty years, one line on this chart looks wrong. Per adult, the Northern Territory’s losses went from around A$900 in the late 1990s to a staggering A$12,976 in 2018–19, about eight times higher than any other state. Then in just one year they dropped back to roughly A$1,400.
People in the NT didn’t suddenly stop gambling. The Territory is where Australia’s big online bookmakers, like Sportsbet and Ladbrokes, are licensed. So for years the whole country’s online betting losses were counted as NT losses. In 2019 the rules changed so bets are taxed where the person actually is, not where the bookie is registered, and all those losses moved away. The spike was never real spending in the NT. Ignore it and the real leader in 2023–24 is New South Wales, with the NT just behind. The chart uses a log scale so the smaller states stay easy to read next to the NT. Click any state in the legend to turn it on or off.
chart4_data <- read_ags(prod_file, "TOTAL 7") %>% filter(state != "Australia")
state_cols <- c(NSW = "#d8352a", NT = "#29339b", VIC = "#b1b1b9", QLD = "#b1b1b9",
SA = "#cfcfd6", ACT = "#cfcfd6", TAS = "#e1e1e3", WA = "#e1e1e3")
chart4_data <- chart4_data %>% mutate(state = factor(state, levels = names(state_cols)))
plot_ly(chart4_data, x = ~year_start, y = ~value, color = ~state, colors = state_cols,
type = "scatter", mode = "lines", line = list(width = 2.5),
hovertemplate = "%{fullData.name}<br>%{x}: A$%{y:.0f}/person<extra></extra>") %>%
conv_layout("Per-person gambling losses by state (click a state to toggle)",
"Expenditure per adult (A$, log scale)") %>%
# log scale here, otherwise NT's massive numbers squash every other state flat
layout(
yaxis = list(type = "log", title = "Expenditure per adult (A$, log scale)",
color = ink, gridcolor = "#f1f1f2"),
annotations = list(
list(x = 2018, y = log10(12976), xref = "x", yref = "y",
text = "NT peak A$12,976/adult:<br>online bookmakers licensed here",
showarrow = TRUE, arrowhead = 2, ax = -60, ay = -30,
font = list(size = 10, color = ink), align = "left")
)
)Figure 4: Total gambling losses per adult by state, 1998–99 to 2023–24 (A$ per adult, log scale). Source: Queensland Government Statistician’s Office (2025).
For state governments, every dollar lost is also a dollar of income. In 2023–24 they collected about A$9.5 billion in gambling taxes, A$3.5 billion of it in NSW alone, followed by Victoria and Queensland.
That money helps pay for hospitals, schools and roads. But it also means the same governments that are meant to reduce gambling harm quietly rely on that harm to keep going. That’s the awkward note this story ends on: the house always wins, and in Australia, the house is partly us.
rev <- read_ags(prod_file, "TOTAL 11") %>% filter(state != "Australia") %>%
mutate(rev_b = value / 1000)
lvl <- rev %>% filter(year_start == 2023) %>% arrange(desc(rev_b)) %>% pull(state)
rev <- rev %>% mutate(state = factor(state, levels = lvl))
plot_ly(rev, x = ~year_start, y = ~rev_b, color = ~state, colors = pal8,
type = "scatter", mode = "lines", stackgroup = "one",
hovertemplate = "%{fullData.name}<br>%{x}: A$%{y:.2f}b<extra></extra>") %>%
conv_layout("Gambling revenue collected by governments, by state",
"Government gambling revenue (A$ billion)")Figure 5: Government gambling revenue by state, over time (nominal A$). Source: Queensland Government Statistician’s Office (2025).
All of the data referenced here are derived from a single source: the Queensland Government Statistician’s Office and its Australian Gambling Statistics series (40th edition). It monitors gambling losses in each state and territory from 1998–99 to 2023–24, with whole national totals available from 1975–76. Except when I specify otherwise, the monetary amounts are nominal, meaning they haven’t been adjusted for inflation. Charts 3 and 4 switch to losses per adult, which lets you compare a big state and a small one fairly.
One figure needs a health warning. The NT’s ridiculously high numbers pre-2019 do not really represent the amount of money lost by people of the Territory. This happens due to the licensing of Australia’s major online bookmakers in the Northern Territory, making the NT responsible for losses from national online gambling until tax reforms in 2019 corrected the issue. Chart 4 keeps those years visible but marks them, instead of pretending they are not there.
I used a generative AI tool (Anthropic’s Claude) for a few things only: tidying the wording, checking my figures against the original spreadsheets, and suggesting the log scale for Chart 4. I did the core work myself: choosing the topic and angle, interpreting the data, shaping the story, and writing all the R code. I checked every AI response against the source data before using it. This is acknowledged in line with the RMIT Library’s guidelines on acknowledging AI.
Thanks too to the Queensland Government Statistician’s Office for putting the data out there for free.
Anthropic. (2026). Claude (Opus 4.8) [Large language model]. https://claude.ai
Field, J. (2017, September 14). Gaming Machines Amendment (Transparency) Bill 2016: Second reading [Hansard]. Parliament of New South Wales. https://www.parliament.nsw.gov.au/bill/files/3352/2R%20Gaming%20Machine.pdf
Queensland Government Statistician’s Office. (2025). Australian gambling statistics, 1998–99 to 2023–24 (40th ed.). Queensland Treasury. https://www.qgso.qld.gov.au/statistics/theme/society/gambling/australian-gambling-statistics